Point mlx-lm to fix/float32-logprobs branch for float32 log_softmax fix

EXO passes repetition_context_size=None to mlx_lm which does
tokens[-None:] causing TypeError. Now automatically sets context
size to 64 when repetition_penalty is provided.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

.github/workflows/build-app.yml

@@ -159,7 +159,7 @@ jobs:
           fi
 
       - name: Install Homebrew packages
-        run: brew install just awscli
+        run: brew install just awscli macmon
 
       - name: Install UV
         uses: astral-sh/setup-uv@v6
@@ -243,14 +243,6 @@ jobs:
       # Build the bundle
       # ============================================================
 
-      - name: Add pinned macmon to PATH
-        run: |
-          MACMON_DIR=$(nix develop --command sh -c 'dirname $(which macmon)')
-          echo "Using macmon from: $MACMON_DIR"
-          echo "$MACMON_DIR" >> $GITHUB_PATH
-          # Remove any Homebrew macmon so PyInstaller can't accidentally pick it up
-          brew uninstall macmon 2>/dev/null || true
-
       - name: Build PyInstaller bundle
         run: uv run pyinstaller packaging/pyinstaller/exo.spec
 

.mlx_typings/mlx/core/__init__.pyi

@@ -2396,7 +2396,7 @@ def degrees(a: array, /, *, stream: Stream | Device | None = ...) -> array:
         array: The angles in degrees.
     """
 
-def depends[T](inputs: T, dependencies: array | Sequence[array]) -> T:
+def depends(inputs: array | Sequence[array], dependencies: array | Sequence[array]):
     """
     Insert dependencies between arrays in the graph. The outputs are
     identical to ``inputs`` but with dependencies on ``dependencies``.

.mlx_typings/mlx/nn/__init__.pyi

@@ -1,5 +1,9 @@
-from .layers import *
-from .utils import *
+"""
+This type stub file was generated by pyright.
+"""
+
+from layers import *
+from utils import *
 
 from . import init as init
 from . import losses as losses

.mlx_typings/mlx/nn/layers/__init__.pyi

@@ -1,16 +1,20 @@
-from .activations import *
-from .base import *
-from .containers import *
-from .convolution import *
-from .convolution_transpose import *
-from .distributed import *
-from .dropout import *
-from .embedding import *
-from .linear import *
-from .normalization import *
-from .pooling import *
-from .positional_encoding import *
-from .quantized import *
-from .recurrent import *
-from .transformer import *
-from .upsample import *
+"""
+This type stub file was generated by pyright.
+"""
+
+from activations import *
+from base import *
+from containers import *
+from convolution import *
+from convolution_transpose import *
+from distributed import *
+from dropout import *
+from embedding import *
+from linear import *
+from normalization import *
+from pooling import *
+from positional_encoding import *
+from quantized import *
+from recurrent import *
+from transformer import *
+from upsample import *

.mlx_typings/mlx/nn/layers/base.pyi

@@ -53,7 +53,7 @@ class Module(dict):
         mx.eval(model.parameters())
     """
 
-    def __call__(self, *args: Any, **kwargs: Any) -> mx.array: ...
+    __call__: Callable
     def __init__(self) -> None:
         """Should be called by the subclasses of ``Module``."""
 

.mlx_typings/mlx/nn/layers/convolution.pyi

@@ -32,7 +32,6 @@ class Conv1d(Module):
     """
 
     weight: mx.array
-    bias: mx.array | None
     groups: int
     def __init__(
         self,

.mlx_typings/mlx/nn/layers/linear.pyi

@@ -40,10 +40,6 @@ class Linear(Module):
         bias (bool, optional): If set to ``False`` then the layer will
           not use a bias. Default is ``True``.
     """
-
-    weight: mx.array
-    bias: mx.array | None
-
     def __init__(self, input_dims: int, output_dims: int, bias: bool = ...) -> None: ...
     def __call__(self, x: mx.array) -> mx.array: ...
     def to_quantized(

.mlx_typings/mlx/nn/layers/normalization.pyi

@@ -88,9 +88,6 @@ class RMSNorm(Module):
         dims (int): The feature dimension of the input to normalize over
         eps (float): A small additive constant for numerical stability
     """
-
-    weight: mx.array
-
     def __init__(self, dims: int, eps: float = ...) -> None: ...
     def __call__(self, x) -> mx.array: ...
 

.mlx_typings/mlx_lm/generate.pyi

@@ -30,7 +30,7 @@ def str2bool(string):  # -> bool:
 def setup_arg_parser():  # -> ArgumentParser:
     """Set up and return the argument parser."""
 
-generation_stream: mx.Stream
+generation_stream = ...
 
 @contextlib.contextmanager
 def wired_limit(
@@ -266,12 +266,12 @@ def _merge_caches(caches: Any) -> List[Any]: ...
 class Batch:
     uids: List[int]
     y: mx.array
-    logprobs: List[mx.array] | mx.array
+    logprobs: mx.array
     max_tokens: List[int]
     num_tokens: List[int]
     cache: List[Any]
-    samplers: List[Callable[[mx.array], mx.array] | None]
-    logits_processors: List[List[Callable[[mx.array, mx.array], mx.array]]]
+    samplers: List[Any]
+    logits_processors: List[Any]
     tokens: List[mx.array]
     def __len__(self) -> int: ...
     def filter(self, keep_idx: List[int]) -> None: ...
@@ -279,18 +279,13 @@ class Batch:
     def extract_cache(self, idx: int) -> List[Any]: ...
 
 class BatchGenerator:
-    model: nn.Module
-    sampler: Callable[[mx.array], mx.array]
-    stop_tokens: set[int]
+    model: Any
     max_kv_size: Optional[int]
     prefill_step_size: int
-    completion_batch_size: int
-    prefill_batch_size: int
     unprocessed_prompts: List[Any]
     active_batch: Optional[Batch]
     prompt_progress_callback: Callable[[List[Tuple[int, int, int]]], None]
     _stats: BatchStats
-    _next_count: int
 
     @dataclass
     class Response:

.mlx_typings/mlx_lm/models/cache.pyi

@@ -88,8 +88,8 @@ def create_attention_mask(
 ) -> array | Literal["causal"] | None: ...
 
 class _BaseCache(Cache):
-    keys: mx.array | None
-    values: mx.array | None
+    keys: mx.array
+    values: mx.array
     offset: int
     @property
     def state(self) -> tuple[mx.array | None, mx.array | None]: ...
@@ -268,14 +268,29 @@ class CacheList(_BaseCache):
         """
 
 class BatchKVCache(_BaseCache):
-    step: int
-    keys: array | None
-    values: array | None
-    offset: array
-    left_padding: array
-    _idx: int
-    def __init__(self, left_padding: List[int]) -> None: ...
-    def update_and_fetch(self, keys: array, values: array) -> tuple[array, array]: ...
+    step = ...
+    def __init__(self, left_padding: List[int]) -> None:
+        """
+        The BatchKV cache expects inputs to be left-padded.
+
+        E.g. the following prompts:
+
+            [1, 3, 5]
+            [7]
+            [2, 6, 8, 9]
+
+        Should be padded like so:
+
+            [0, 1, 3, 5]
+            [0, 0, 0, 7]
+            [2, 6, 8, 9]
+
+        And ``left_padding`` specifies the amount of padding for each.
+        In this case, ``left_padding = [1, 3, 0]``.
+        """
+
+    def update_and_fetch(self, keys, values):  # -> tuple[array | Any, array | Any]:
+        ...
     @property
     def state(
         self,
@@ -301,21 +316,12 @@ class BatchKVCache(_BaseCache):
         """
 
 class BatchRotatingKVCache(_BaseCache):
-    step: int
-    keys: array | None
-    values: array | None
-    offset: array
-    left_padding: array
-    max_size: int
-    _idx: int
-    _offset: int
-    rotated: bool
-    _lengths: array | None
-    def __init__(self, max_size: int, left_padding: List[int]) -> None: ...
-    def _trim(self, trim_size: int, v: array, append: array | None = ...) -> array: ...
-    def _update_in_place(self, keys: array, values: array) -> tuple[array, array]: ...
-    def _update_concat(self, keys: array, values: array) -> tuple[array, array]: ...
-    def update_and_fetch(self, keys: array, values: array) -> tuple[array, array]: ...
+    step = ...
+    def __init__(self, max_size, left_padding: List[int]) -> None: ...
+    def update_and_fetch(
+        self, keys, values
+    ):  # -> tuple[array | Any, array | Any] | tuple[array | Any, array | Any | None]:
+        ...
     @property
     def state(
         self,

.mlx_typings/mlx_lm/models/gated_delta.pyi

@@ -1,35 +0,0 @@
-from typing import Optional
-
-import mlx.core as mx
-
-def compute_g(A_log: mx.array, a: mx.array, dt_bias: mx.array) -> mx.array: ...
-def gated_delta_update(
-    q: mx.array,
-    k: mx.array,
-    v: mx.array,
-    a: mx.array,
-    b: mx.array,
-    A_log: mx.array,
-    dt_bias: mx.array,
-    state: Optional[mx.array] = ...,
-    mask: Optional[mx.array] = ...,
-    use_kernel: bool = ...,
-) -> tuple[mx.array, mx.array]: ...
-def gated_delta_ops(
-    q: mx.array,
-    k: mx.array,
-    v: mx.array,
-    g: mx.array,
-    beta: mx.array,
-    state: Optional[mx.array] = ...,
-    mask: Optional[mx.array] = ...,
-) -> tuple[mx.array, mx.array]: ...
-def gated_delta_kernel(
-    q: mx.array,
-    k: mx.array,
-    v: mx.array,
-    g: mx.array,
-    beta: mx.array,
-    state: mx.array,
-    mask: Optional[mx.array] = ...,
-) -> tuple[mx.array, mx.array]: ...

.mlx_typings/mlx_lm/models/nemotron_h.pyi

@@ -1,154 +0,0 @@
-from dataclasses import dataclass
-from typing import Any, List, Optional, Tuple
-
-import mlx.core as mx
-import mlx.nn as nn
-
-from .cache import ArraysCache, KVCache
-from .switch_layers import SwitchMLP
-
-@dataclass
-class ModelArgs:
-    model_type: str
-    vocab_size: int
-    hidden_size: int
-    intermediate_size: int
-    num_hidden_layers: int
-    max_position_embeddings: int
-    num_attention_heads: int
-    num_key_value_heads: int
-    attention_bias: bool
-    mamba_num_heads: int
-    mamba_head_dim: int
-    mamba_proj_bias: bool
-    ssm_state_size: int
-    conv_kernel: int
-    n_groups: int
-    mlp_bias: bool
-    layer_norm_epsilon: float
-    use_bias: bool
-    use_conv_bias: bool
-    hybrid_override_pattern: List[str]
-    head_dim: Optional[int]
-    moe_intermediate_size: Optional[int]
-    moe_shared_expert_intermediate_size: Optional[int]
-    n_group: Optional[int]
-    n_routed_experts: Optional[int]
-    n_shared_experts: Optional[int]
-    topk_group: Optional[int]
-    num_experts_per_tok: Optional[int]
-    norm_topk_prob: Optional[bool]
-    routed_scaling_factor: Optional[float]
-    time_step_limit: Optional[Tuple[float, float]]
-    time_step_min: Optional[float]
-    time_step_max: Optional[float]
-
-    @classmethod
-    def from_dict(cls, params: dict[str, Any]) -> ModelArgs: ...
-    def __post_init__(self) -> None: ...
-
-class NemotronHMamba2Mixer(nn.Module):
-    num_heads: int
-    hidden_size: int
-    ssm_state_size: int
-    conv_kernel_size: int
-    intermediate_size: int
-    n_groups: int
-    head_dim: int
-    conv_dim: int
-    conv1d: nn.Conv1d
-    in_proj: nn.Linear
-    dt_bias: mx.array
-    A_log: mx.array
-    D: mx.array
-    norm: nn.RMSNorm
-    heads_per_group: int
-    out_proj: nn.Linear
-
-    def __init__(self, args: ModelArgs) -> None: ...
-    def __call__(
-        self,
-        hidden_states: mx.array,
-        mask: Optional[mx.array],
-        cache: Optional[ArraysCache] = None,
-    ) -> mx.array: ...
-
-class NemotronHAttention(nn.Module):
-    hidden_size: int
-    num_heads: int
-    head_dim: int
-    num_key_value_heads: int
-    scale: float
-    q_proj: nn.Linear
-    k_proj: nn.Linear
-    v_proj: nn.Linear
-    o_proj: nn.Linear
-
-    def __init__(self, args: ModelArgs) -> None: ...
-    def __call__(
-        self,
-        x: mx.array,
-        mask: Optional[mx.array] = None,
-        cache: Optional[KVCache] = None,
-    ) -> mx.array: ...
-
-class NemotronHMLP(nn.Module):
-    up_proj: nn.Linear
-    down_proj: nn.Linear
-
-    def __init__(
-        self, args: ModelArgs, intermediate_size: Optional[int] = None
-    ) -> None: ...
-    def __call__(self, x: mx.array) -> mx.array: ...
-
-class NemotronHMoE(nn.Module):
-    num_experts_per_tok: int
-    switch_mlp: SwitchMLP
-    shared_experts: NemotronHMLP
-
-    def __init__(self, config: ModelArgs) -> None: ...
-    def __call__(self, x: mx.array) -> mx.array: ...
-
-class NemotronHBlock(nn.Module):
-    block_type: str
-    norm: nn.RMSNorm
-    mixer: NemotronHMamba2Mixer | NemotronHAttention | NemotronHMLP | NemotronHMoE
-
-    def __init__(self, args: ModelArgs, block_type: str) -> None: ...
-    def __call__(
-        self,
-        x: mx.array,
-        mask: Optional[mx.array] = None,
-        cache: Optional[Any] = None,
-    ) -> mx.array: ...
-
-class NemotronHModel(nn.Module):
-    embeddings: nn.Embedding
-    layers: list[NemotronHBlock]
-    norm_f: nn.RMSNorm
-    fa_idx: int
-    ssm_idx: int
-
-    def __init__(self, args: ModelArgs) -> None: ...
-    def __call__(
-        self,
-        inputs: mx.array,
-        cache: Optional[Any] = None,
-    ) -> mx.array: ...
-
-class Model(nn.Module):
-    args: ModelArgs
-    backbone: NemotronHModel
-    lm_head: nn.Linear
-    model_type: str
-
-    def __init__(self, args: ModelArgs) -> None: ...
-    def __call__(
-        self,
-        inputs: mx.array,
-        cache: Optional[Any] = None,
-    ) -> mx.array: ...
-    @property
-    def layers(self) -> list[NemotronHBlock]: ...
-    def make_cache(self) -> list[ArraysCache | KVCache]: ...
-    def sanitize(self, weights: dict[str, Any]) -> dict[str, Any]: ...

.mlx_typings/mlx_lm/models/qwen3_next.pyi

@@ -5,7 +5,6 @@ from typing import Any, Optional
 import mlx.core as mx
 import mlx.nn as nn
 
-from .cache import ArraysCache, KVCache
 from .switch_layers import SwitchGLU
 
 class Qwen3NextRMSNormGated(nn.Module):
@@ -100,8 +99,6 @@ class Qwen3NextModel(nn.Module):
     embed_tokens: nn.Embedding
     layers: list[Qwen3NextDecoderLayer]
     norm: nn.RMSNorm
-    ssm_idx: int
-    fa_idx: int
 
     def __init__(self, args: Any) -> None: ...
     def __call__(
@@ -124,4 +121,3 @@ class Model(nn.Module):
     def sanitize(self, weights: dict[str, Any]) -> dict[str, Any]: ...
     @property
     def layers(self) -> list[Qwen3NextDecoderLayer]: ...
-    def make_cache(self) -> list[ArraysCache | KVCache]: ...

.mlx_typings/mlx_lm/models/rope_utils.pyi

@@ -1,51 +0,0 @@
-from typing import Any, Optional
-
-import mlx.nn as nn
-
-class YarnRoPE(nn.Module):
-    def __init__(
-        self,
-        dims: int,
-        traditional: bool = ...,
-        max_position_embeddings: int = ...,
-        base: float = ...,
-        scaling_factor: float = ...,
-        original_max_position_embeddings: int = ...,
-        beta_fast: float = ...,
-        beta_slow: float = ...,
-        mscale: float = ...,
-        mscale_all_dim: float = ...,
-    ) -> None: ...
-
-class Llama3RoPE(nn.Module):
-    def __init__(
-        self,
-        dims: int,
-        traditional: bool = ...,
-        max_position_embeddings: int = ...,
-        base: float = ...,
-        scaling_factor: float = ...,
-        original_max_position_embeddings: int = ...,
-        low_freq_factor: float = ...,
-        high_freq_factor: float = ...,
-    ) -> None: ...
-
-class SuScaledRoPE(nn.Module):
-    def __init__(
-        self,
-        dims: int,
-        traditional: bool = ...,
-        max_position_embeddings: int = ...,
-        base: float = ...,
-        short_factor: Any = ...,
-        long_factor: Any = ...,
-        original_max_position_embeddings: int = ...,
-    ) -> None: ...
-
-def initialize_rope(
-    dims: int,
-    base: float = ...,
-    traditional: bool = ...,
-    scaling_config: Optional[dict[str, Any]] = ...,
-    max_position_embeddings: Optional[int] = ...,
-) -> nn.Module: ...

.mlx_typings/mlx_lm/models/switch_layers.pyi

@@ -73,9 +73,6 @@ class SwitchGLU(nn.Module):
     def __call__(self, x, indices) -> mx.array: ...
 
 class SwitchMLP(nn.Module):
-    fc1: SwitchLinear
-    fc2: SwitchLinear
-
     def __init__(
         self,
         input_dims: int,

.mlx_typings/mlx_lm/tokenizer_utils.pyi

@@ -39,11 +39,11 @@ class StreamingDetokenizer:
     """
 
     __slots__ = ...
-    def reset(self) -> None: ...
-    def add_token(self, token: int) -> None: ...
-    def finalize(self) -> None: ...
+    def reset(self): ...
+    def add_token(self, token): ...
+    def finalize(self): ...
     @property
-    def last_segment(self) -> str:
+    def last_segment(self):
         """Return the last segment of readable text since last time this property was accessed."""
 
 class NaiveStreamingDetokenizer(StreamingDetokenizer):

.mlx_typings/mlx_vlm/prompt_utils.pyi

@@ -1,12 +0,0 @@
-from typing import Any
-
-def get_message_json(
-    model_name: str,
-    prompt: str,
-    role: str = "user",
-    skip_image_token: bool = False,
-    skip_audio_token: bool = False,
-    num_images: int = 0,
-    num_audios: int = 0,
-    **kwargs: Any,
-) -> dict[str, Any]: ...

.mlx_typings/mlx_vlm/utils.pyi

@@ -1,15 +0,0 @@
-from pathlib import Path
-from typing import Any
-
-class ImageProcessor:
-    def preprocess(
-        self, images: list[dict[str, Any]], **kwargs: Any
-    ) -> dict[str, Any]: ...
-    def __call__(self, **kwargs: Any) -> dict[str, Any]: ...
-
-def load_image_processor(
-    model_path: str | Path, **kwargs: Any
-) -> ImageProcessor | None: ...
-def load_processor(
-    model_path: str | Path, add_detokenizer: bool = ..., **kwargs: Any
-) -> ImageProcessor: ...

.mlx_typings/safetensors/__init__.pyi

@@ -1,8 +0,0 @@
-from typing import Any, Self
-
-class safe_open:
-    def __init__(self, filename: str, framework: str = "pt") -> None: ...
-    def __enter__(self) -> Self: ...
-    def __exit__(self, *args: Any) -> None: ...
-    def keys(self) -> list[str]: ...
-    def get_tensor(self, name: str) -> Any: ...

CONTRIBUTING.md

@@ -11,18 +11,9 @@ To run EXO from source:
   ```bash
   brew install uv
   ```
-- [rust](https://github.com/rust-lang/rustup) (to build Rust bindings, nightly for now)
-  ```bash
-  curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
-  rustup toolchain install nightly
-  ```
 - [macmon](https://github.com/vladkens/macmon) (for hardware monitoring on Apple Silicon)
-  Use the pinned fork revision used by this repo instead of Homebrew `macmon`.
   ```bash
-  cargo install --git https://github.com/swiftraccoon/macmon \
-    --rev 9154d234f763fbeffdcb4135d0bbbaf80609699b \
-    macmon \
-    --force
+  brew install macmon
   ```
 
 ```bash

README.md

@@ -95,10 +95,11 @@ Then restart the Nix daemon: `sudo launchctl kickstart -k system/org.nixos.nix-d
   /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
   ```
 - [uv](https://github.com/astral-sh/uv) (for Python dependency management)
+- [macmon](https://github.com/vladkens/macmon) (for hardware monitoring on Apple Silicon)
 - [node](https://github.com/nodejs/node) (for building the dashboard)
 
   ```bash
-  brew install uv node
+  brew install uv macmon node
   ```
 - [rust](https://github.com/rust-lang/rustup) (to build Rust bindings, nightly for now)
 
@@ -106,17 +107,6 @@ Then restart the Nix daemon: `sudo launchctl kickstart -k system/org.nixos.nix-d
   curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
   rustup toolchain install nightly
   ```
-- [macmon](https://github.com/vladkens/macmon) (for hardware monitoring on Apple Silicon)
-
-  Install the pinned fork revision used by this repo instead of Homebrew `macmon`.
-  Homebrew `macmon 0.6.1` still crashes on Apple M5.
-
-  ```bash
-  cargo install --git https://github.com/swiftraccoon/macmon \
-    --rev 9154d234f763fbeffdcb4135d0bbbaf80609699b \
-    macmon \
-    --force
-  ```
 
 Clone the repo, build the dashboard, and run exo:
 
@@ -295,9 +285,8 @@ exo supports several environment variables for configuration:
 
 | Variable | Description | Default |
 |----------|-------------|---------|
-| `EXO_DEFAULT_MODELS_DIR` | Default directory for model downloads and caches. Always first in the writable dirs list. | `~/.local/share/exo/models` (Linux) or `~/.exo/models` (macOS) |
-| `EXO_MODELS_DIRS` | Colon-separated additional writable directories for model downloads. Checked in order after the default; first with enough free space is used. | None |
-| `EXO_MODELS_READ_ONLY_DIRS` | Colon-separated read-only directories to search for pre-downloaded models (e.g., NFS mounts, shared storage). Models here cannot be deleted. | None |
+| `EXO_MODELS_PATH` | Colon-separated paths to search for pre-downloaded models (e.g., on NFS mounts or shared storage) | None |
+| `EXO_MODELS_DIR` | Directory where exo downloads and stores models | `~/.local/share/exo/models` (Linux) or `~/.exo/models` (macOS) |
 | `EXO_OFFLINE` | Run without internet connection (uses only local models) | `false` |
 | `EXO_ENABLE_IMAGE_MODELS` | Enable image model support | `false` |
 | `EXO_LIBP2P_NAMESPACE` | Custom namespace for cluster isolation | None |
@@ -307,11 +296,8 @@ exo supports several environment variables for configuration:
 **Example usage:**
 
 ```bash
-# Use pre-downloaded models from NFS mount (read-only)
-EXO_MODELS_READ_ONLY_DIRS=/mnt/nfs/models:/opt/ai-models uv run exo
-
-# Download models to an external SSD (falls back to default dir if full)
-EXO_MODELS_DIRS=/Volumes/ExternalSSD/exo-models uv run exo
+# Use pre-downloaded models from NFS mount
+EXO_MODELS_PATH=/mnt/nfs/models:/opt/ai-models uv run exo
 
 # Run in offline mode
 EXO_OFFLINE=true uv run exo

app/EXO/EXO/ExoProcessController.swift

@@ -4,7 +4,6 @@ import Foundation
 
 private let customNamespaceKey = "EXOCustomNamespace"
 private let hfTokenKey = "EXOHFToken"
-private let hfEndpointKey = "EXOHFEndpoint"
 private let enableImageModelsKey = "EXOEnableImageModels"
 private let offlineModeKey = "EXOOfflineMode"
 private let onboardingCompletedKey = "EXOOnboardingCompleted"
@@ -54,14 +53,6 @@ final class ExoProcessController: ObservableObject {
             UserDefaults.standard.set(hfToken, forKey: hfTokenKey)
         }
     }
-    @Published var hfEndpoint: String = {
-        return UserDefaults.standard.string(forKey: hfEndpointKey) ?? ""
-    }()
-    {
-        didSet {
-            UserDefaults.standard.set(hfEndpoint, forKey: hfEndpointKey)
-        }
-    }
     @Published var enableImageModels: Bool = {
         return UserDefaults.standard.bool(forKey: enableImageModelsKey)
     }()
@@ -282,9 +273,6 @@ final class ExoProcessController: ObservableObject {
         if !hfToken.isEmpty {
             environment["HF_TOKEN"] = hfToken
         }
-        if !hfEndpoint.isEmpty {
-            environment["HF_ENDPOINT"] = hfEndpoint
-        }
         if enableImageModels {
             environment["EXO_ENABLE_IMAGE_MODELS"] = "true"
         }

app/EXO/EXO/Views/SettingsView.swift

@@ -12,7 +12,6 @@ struct SettingsView: View {
 
     @State private var pendingNamespace: String = ""
     @State private var pendingHFToken: String = ""
-    @State private var pendingHFEndpoint: String = ""
     @State private var pendingEnableImageModels = false
     @State private var pendingOfflineMode = false
     @State private var needsRestart = false
@@ -43,7 +42,6 @@ struct SettingsView: View {
         .onAppear {
             pendingNamespace = controller.customNamespace
             pendingHFToken = controller.hfToken
-            pendingHFEndpoint = controller.hfEndpoint
             pendingEnableImageModels = controller.enableImageModels
             pendingOfflineMode = controller.offlineMode
             needsRestart = false
@@ -76,17 +74,6 @@ struct SettingsView: View {
                     .foregroundColor(.secondary)
             }
 
-            Section {
-                LabeledContent("HuggingFace Endpoint") {
-                    TextField("default", text: $pendingHFEndpoint)
-                        .textFieldStyle(.roundedBorder)
-                        .frame(width: 200)
-                }
-                Text("Defaults to huggingface.co. Use a mirror (e.g. hf-mirror.com) for China.")
-                    .font(.caption)
-                    .foregroundColor(.secondary)
-            }
-
             Section {
                 Toggle("Offline Mode", isOn: $pendingOfflineMode)
                 Text("Skip internet checks and use only locally available models.")
@@ -467,7 +454,6 @@ struct SettingsView: View {
 
     private var hasGeneralChanges: Bool {
         pendingNamespace != controller.customNamespace || pendingHFToken != controller.hfToken
-            || pendingHFEndpoint != controller.hfEndpoint
             || pendingOfflineMode != controller.offlineMode
     }
 
@@ -478,7 +464,6 @@ struct SettingsView: View {
     private func applyGeneralSettings() {
         controller.customNamespace = pendingNamespace
         controller.hfToken = pendingHFToken
-        controller.hfEndpoint = pendingHFEndpoint
         controller.offlineMode = pendingOfflineMode
         restartIfRunning()
     }

bench/eval_configs/models.toml

@@ -7,7 +7,7 @@
 #   name, patterns, reasoning
 #
 # Optional per-model overrides (CLI flags take priority over these):
-#   temperature, top_p, max_tokens, reasoning_effort, enable_thinking
+#   temperature, top_p, max_tokens, reasoning_effort
 #
 # Fallback defaults (when no per-model config):
 #   reasoning:     temperature=1.0, max_tokens=131072, reasoning_effort="high"
@@ -18,9 +18,10 @@
 
 # ─── Qwen3.5 (Feb 2026) ─────────────────────────────────────────────
 # Source: HuggingFace model cards (Qwen/Qwen3.5-*)
-# Model card recommends: temp=0.6, top_p=0.95, top_k=20
-# We omit top_k to match vllm eval (which doesn't set it).
-# max_tokens=121072 to match vllm eval (131072 context - 10000 safety margin).
+# 35B-A3B thinking general: temp=1.0, top_p=0.95, top_k=20
+# 397B thinking: temp=0.6, top_p=0.95, top_k=20
+# Non-thinking: temp=0.7, top_p=0.8, top_k=20
+# max_tokens: 32768 general, 81920 for complex math/code
 
 [[model]]
 name = "Qwen3.5 2B"
@@ -28,8 +29,7 @@ patterns = ["Qwen3.5-2B"]
 reasoning = true
 temperature = 0.6
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
+max_tokens = 81920
 
 [[model]]
 name = "Qwen3.5 9B"
@@ -37,8 +37,7 @@ patterns = ["Qwen3.5-9B"]
 reasoning = true
 temperature = 0.6
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
+max_tokens = 81920
 
 [[model]]
 name = "Qwen3.5 27B"
@@ -46,17 +45,15 @@ patterns = ["Qwen3.5-27B"]
 reasoning = true
 temperature = 0.6
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
+max_tokens = 81920
 
 [[model]]
 name = "Qwen3.5 35B A3B"
 patterns = ["Qwen3.5-35B-A3B"]
 reasoning = true
-temperature = 0.6
+temperature = 1.0
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
+max_tokens = 81920
 
 [[model]]
 name = "Qwen3.5 122B A10B"
@@ -64,8 +61,7 @@ patterns = ["Qwen3.5-122B-A10B"]
 reasoning = true
 temperature = 0.6
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
+max_tokens = 81920
 
 [[model]]
 name = "Qwen3.5 397B A17B"
@@ -73,14 +69,12 @@ patterns = ["Qwen3.5-397B-A17B"]
 reasoning = true
 temperature = 0.6
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
+max_tokens = 81920
 
 # ─── Qwen3 (Apr 2025) ───────────────────────────────────────────────
 # Source: HuggingFace model cards (Qwen/Qwen3-*)
-# Model card recommends: temp=0.6, top_p=0.95, top_k=20
-# We omit top_k to match vllm eval (which doesn't set it).
-# Non-thinking: temp=0.7, top_p=0.8
+# Thinking: temp=0.6, top_p=0.95, top_k=20
+# Non-thinking: temp=0.7, top_p=0.8, top_k=20
 # max_tokens: 32768 general, 38912 for complex math/code
 
 [[model]]
@@ -89,7 +83,6 @@ patterns = ["Qwen3-0.6B"]
 reasoning = true
 temperature = 0.6
 top_p = 0.95
-enable_thinking = true
 max_tokens = 38912
 
 [[model]]
@@ -98,7 +91,6 @@ patterns = ["Qwen3-30B-A3B"]
 reasoning = true
 temperature = 0.6
 top_p = 0.95
-enable_thinking = true
 max_tokens = 38912
 
 [[model]]
@@ -107,7 +99,6 @@ patterns = ["Qwen3-235B-A22B"]
 reasoning = true
 temperature = 0.6
 top_p = 0.95
-enable_thinking = true
 max_tokens = 38912
 
 [[model]]
@@ -116,7 +107,6 @@ patterns = ["Qwen3-Next-80B-A3B-Thinking"]
 reasoning = true
 temperature = 0.6
 top_p = 0.95
-enable_thinking = true
 max_tokens = 38912
 
 [[model]]
@@ -139,9 +129,9 @@ max_tokens = 16384
 name = "Qwen3 Coder Next"
 patterns = ["Qwen3-Coder-Next"]
 reasoning = false
-temperature = 1.0
-top_p = 0.95
-max_tokens = 121072
+temperature = 0.7
+top_p = 0.8
+max_tokens = 16384
 
 # ─── GPT-OSS (OpenAI) ───────────────────────────────────────────────
 # Source: OpenAI GitHub README + HuggingFace discussion #21
@@ -175,38 +165,10 @@ patterns = ["DeepSeek-V3.1"]
 reasoning = true
 temperature = 0.0
 
-[[model]]
-name = "DeepSeek V3.2"
-patterns = ["DeepSeek-V3.2"]
-reasoning = true
-temperature = 1.0
-top_p = 0.95
-enable_thinking = true
-
-# ─── NVIDIA Nemotron ───────────────────────────────────────────────────
-# Source: HuggingFace model cards
-# All variants: temp=1.0, top_p=0.95, enable_thinking=true
-
-[[model]]
-name = "Nemotron Cascade 2 30B A3B"
-patterns = ["Nemotron-Cascade-2-30B-A3B"]
-reasoning = true
-temperature = 1.0
-top_p = 0.95
-enable_thinking = true
-
-[[model]]
-name = "Nemotron 3 Super 120B A12B"
-patterns = ["Nemotron-3-Super-120B-A12B", "NVIDIA-Nemotron-3-Super-120B-A12B"]
-reasoning = true
-temperature = 1.0
-top_p = 0.95
-enable_thinking = true
-
 # ─── GLM (ZhipuAI / THUDM) ──────────────────────────────────────────
 # Source: HuggingFace model cards + generation_config.json + docs.z.ai
 # GLM 4.5+: temp=1.0, top_p=0.95
-# max_tokens=121072 to match vllm eval (131072 context - 10000 safety margin)
+# Reasoning tasks: 131072 max_tokens; coding/SWE tasks: temp=0.7
 
 [[model]]
 name = "GLM-5"
@@ -214,8 +176,7 @@ patterns = ["GLM-5"]
 reasoning = true
 temperature = 1.0
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
+max_tokens = 131072
 
 [[model]]
 name = "GLM 4.5 Air"
@@ -230,8 +191,7 @@ patterns = ["GLM-4.7-"]
 reasoning = true
 temperature = 1.0
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
+max_tokens = 131072
 # Note: matches both GLM-4.7 and GLM-4.7-Flash
 
 # ─── Kimi (Moonshot AI) ─────────────────────────────────────────────
@@ -253,8 +213,7 @@ patterns = ["Kimi-K2.5"]
 reasoning = true
 temperature = 1.0
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
+max_tokens = 131072
 
 [[model]]
 name = "Kimi K2 Instruct"
@@ -264,8 +223,7 @@ temperature = 0.6
 
 # ─── MiniMax ─────────────────────────────────────────────────────────
 # Source: HuggingFace model cards + generation_config.json
-# All models: temp=1.0, top_p=0.95
-# max_tokens=90000 to match vllm eval (100000 context - 10000 safety margin)
+# All models: temp=1.0, top_p=0.95, top_k=40
 
 [[model]]
 name = "MiniMax M2.5"
@@ -273,8 +231,6 @@ patterns = ["MiniMax-M2.5"]
 reasoning = true
 temperature = 1.0
 top_p = 0.95
-enable_thinking = true
-max_tokens = 90000
 
 [[model]]
 name = "MiniMax M2.1"
@@ -295,8 +251,6 @@ patterns = ["Step-3.5-Flash"]
 reasoning = true
 temperature = 1.0
 top_p = 0.95
-enable_thinking = true
-max_tokens = 121072
 
 # ─── Llama (Meta) ───────────────────────────────────────────────────
 # Source: generation_config.json + meta-llama/llama-models generation.py

bench/eval_tool_calls.py

@@ -17,7 +17,6 @@ from harness import (
     ExoClient,
     ExoHttpError,
     add_common_instance_args,
-    capture_cluster_snapshot,
     instance_id_from_instance,
     nodes_used_in_instance,
     resolve_model_short_id,
@@ -1007,7 +1006,6 @@ Examples:
         sys.exit(1)
 
     time.sleep(1)
-    cluster_snapshot = capture_cluster_snapshot(exo)
     all_results: list[ScenarioResult] = []
 
     try:
@@ -1086,19 +1084,16 @@ Examples:
                 print(f"  - {r.name} [{r.api}/{r.phase}]: {r.error}", file=log)
 
     json_results = [result_to_dict(r) for r in all_results]
-    output: dict[str, Any] = {"results": json_results}
-    if cluster_snapshot:
-        output["cluster"] = cluster_snapshot
 
     if args.stdout:
-        print(json.dumps(output, indent=2))
+        print(json.dumps(json_results, indent=2))
     else:
         json_path = args.json_out
         parent = os.path.dirname(json_path)
         if parent:
             os.makedirs(parent, exist_ok=True)
         with open(json_path, "w") as f:
-            json.dump(output, f, indent=2)
+            json.dump(json_results, f, indent=2)
             f.write("\n")
         print(f"\nJSON results written to {json_path}", file=log)
 

bench/exo_bench.py

@@ -22,7 +22,6 @@ import contextlib
 import itertools
 import json
 import sys
-import threading
 import time
 from collections.abc import Callable
 from concurrent.futures import ThreadPoolExecutor, as_completed
@@ -34,9 +33,7 @@ from harness import (
     ExoClient,
     ExoHttpError,
     add_common_instance_args,
-    capture_cluster_snapshot,
     instance_id_from_instance,
-    node_ids_from_instance,
     nodes_used_in_instance,
     resolve_model_short_id,
     run_planning_phase,
@@ -79,7 +76,7 @@ def load_tokenizer_for_bench(model_id: str) -> Any:
         model_path = Path(
             snapshot_download(
                 model_id,
-                allow_patterns=["*.json", "*.py", "*.tiktoken", "*.model", "*.jinja"],
+                allow_patterns=["*.json", "*.py", "*.tiktoken", "*.model"],
             )
         )
 
@@ -134,91 +131,6 @@ def format_peak_memory(b: float) -> str:
     raise ValueError("You're using petabytes of memory. Something went wrong...")
 
 
-_SAMPLER_METRICS = ("gpuUsage", "temp", "sysPower", "pcpuUsage", "ecpuUsage")
-
-
-class SystemMetricsSampler:
-    def __init__(self, client: ExoClient, node_ids: list[str], interval_s: float = 1.0):
-        self._client = client
-        self._node_ids = node_ids
-        self._interval_s = interval_s
-        self._samples: dict[str, list[tuple[float, dict[str, float]]]] = {
-            nid: [] for nid in node_ids
-        }
-        self._stop = threading.Event()
-        self._thread: threading.Thread | None = None
-
-    def start(self) -> None:
-        self._stop.clear()
-        self._thread = threading.Thread(target=self._poll_loop, daemon=True)
-        self._thread.start()
-
-    def stop(self) -> None:
-        self._stop.set()
-        if self._thread:
-            self._thread.join(timeout=5)
-
-    def _poll_loop(self) -> None:
-        while not self._stop.is_set():
-            t = time.monotonic()
-            for nid in self._node_ids:
-                try:
-                    data = self._client.get_node_system(nid)
-                    if data:
-                        self._samples[nid].append(
-                            (t, {k: data.get(k, 0.0) for k in _SAMPLER_METRICS})
-                        )
-                except Exception:
-                    pass
-            self._stop.wait(self._interval_s)
-
-    def energy_between(self, t0: float, t1: float) -> float:
-        total_joules = 0.0
-        for _nid, samples in self._samples.items():
-            window = [(t, s["sysPower"]) for t, s in samples if t0 <= t <= t1]
-            if len(window) >= 2:
-                for i in range(1, len(window)):
-                    dt = window[i][0] - window[i - 1][0]
-                    avg_power = (window[i][1] + window[i - 1][1]) / 2
-                    total_joules += avg_power * dt
-            elif len(window) == 1:
-                total_joules += window[0][1] * (t1 - t0)
-        return total_joules
-
-    def summarize(self) -> dict[str, dict[str, dict[str, float]]]:
-        result: dict[str, dict[str, dict[str, float]]] = {}
-        for nid, samples in self._samples.items():
-            if not samples:
-                continue
-            metrics: dict[str, dict[str, float]] = {}
-            for key in _SAMPLER_METRICS:
-                values = [s[key] for t, s in samples]
-                metrics[key] = {
-                    "min": round(min(values), 2),
-                    "max": round(max(values), 2),
-                    "mean": round(mean(values), 2),
-                    "samples": len(values),
-                }
-            result[nid] = metrics
-        return result
-
-    def print_summary(self, placement_label: str) -> None:
-        summary = self.summarize()
-        if not summary:
-            return
-        logger.info(f"--- System Metrics ({placement_label}) ---")
-        for nid, metrics in summary.items():
-            gpu = metrics.get("gpuUsage", {})
-            temp = metrics.get("temp", {})
-            power = metrics.get("sysPower", {})
-            logger.info(
-                f"  {nid}:  "
-                f"GPU {gpu.get('mean', 0) * 100:.0f}% avg ({gpu.get('min', 0) * 100:.0f}–{gpu.get('max', 0) * 100:.0f}%)  |  "
-                f"{temp.get('mean', 0):.1f}°C avg  |  "
-                f"{power.get('mean', 0):.1f}W avg"
-            )
-
-
 def parse_int_list(values: list[str]) -> list[int]:
     items: list[int] = []
     for v in values:
@@ -367,17 +279,6 @@ def main() -> int:
         action="store_true",
         help="Force all pp×tg combinations (cartesian product) even when lists have equal length.",
     )
-    ap.add_argument(
-        "--no-system-metrics",
-        action="store_true",
-        help="Disable GPU utilization, temperature, and power collection during inference.",
-    )
-    ap.add_argument(
-        "--metrics-interval",
-        type=float,
-        default=1.0,
-        help="System metrics polling interval in seconds (default: 1.0).",
-    )
     args = ap.parse_args()
 
     pp_list = parse_int_list(args.pp)
@@ -430,7 +331,7 @@ def main() -> int:
         key=lambda p: (
             str(p.get("instance_meta", "")),
             str(p.get("sharding", "")),
-            nodes_used_in_instance(p["instance"]),
+            -nodes_used_in_instance(p["instance"]),
         ),
         reverse=True,
     )
@@ -463,9 +364,7 @@ def main() -> int:
     else:
         logger.info("Download: model already cached")
 
-    cluster_snapshot = capture_cluster_snapshot(client)
     all_rows: list[dict[str, Any]] = []
-    all_system_metrics: dict[str, dict[str, dict[str, float]]] = {}
 
     for preview in selected:
         instance = preview["instance"]
@@ -491,16 +390,6 @@ def main() -> int:
 
         time.sleep(1)
 
-        sampler: SystemMetricsSampler | None = None
-        if not args.no_system_metrics:
-            nids = node_ids_from_instance(instance)
-            sampler = SystemMetricsSampler(
-                ExoClient(args.host, args.port, timeout_s=30),
-                nids,
-                interval_s=args.metrics_interval,
-            )
-            sampler.start()
-
         try:
             for i in range(args.warmup):
                 run_one_completion(
@@ -519,18 +408,15 @@ def main() -> int:
                 for concurrency in concurrency_list:
                     logger.info(f"--- pp={pp} tg={tg} concurrency={concurrency} ---")
                     runs: list[dict[str, Any]] = []
-                    inference_windows: list[tuple[float, float]] = []
                     for r in range(args.repeat):
                         time.sleep(3)
 
                         if concurrency <= 1:
                             # Sequential: single request
                             try:
-                                inf_t0 = time.monotonic()
                                 row, actual_pp_tokens = run_one_completion(
                                     client, full_model_id, pp, tg, prompt_sizer
                                 )
-                                inference_windows.append((inf_t0, time.monotonic()))
                             except Exception as e:
                                 logger.error(e)
                                 continue
@@ -571,7 +457,6 @@ def main() -> int:
                                     c, full_model_id, _pp, _tg, prompt_sizer
                                 )
 
-                            inf_t0 = time.monotonic()
                             with ThreadPoolExecutor(max_workers=concurrency) as pool:
                                 futures = {
                                     pool.submit(_run_concurrent, i): i
@@ -583,7 +468,6 @@ def main() -> int:
                                     except Exception as e:
                                         logger.error(f"Concurrent request failed: {e}")
                                         batch_errors += 1
-                            inference_windows.append((inf_t0, time.monotonic()))
 
                             for idx, (row, actual_pp_tokens) in enumerate(
                                 batch_results
@@ -612,55 +496,33 @@ def main() -> int:
                                 all_rows.append(row)
 
                             if batch_results:
-                                valid_gen_tps = [
+                                agg_gen_tps = sum(
                                     x["stats"]["generation_tps"]
                                     for x, _ in batch_results
                                     if x["stats"]["generation_tps"] > 0
-                                ]
-                                per_req_tps = (
-                                    mean(valid_gen_tps) if valid_gen_tps else 0.0
                                 )
-                                agg_gen_tps = per_req_tps * concurrency
                                 logger.info(
                                     f"[concurrent {concurrency}x]  "
                                     f"agg_gen_tps={agg_gen_tps:.2f}  "
-                                    f"per_req_tps={per_req_tps:.2f}  "
                                     f"errors={batch_errors}"
                                 )
 
                     if runs:
                         prompt_tps = mean(x["stats"]["prompt_tps"] for x in runs)
-                        per_req_tps = mean(x["stats"]["generation_tps"] for x in runs)
-                        gen_tps = per_req_tps * concurrency
+                        gen_tps = mean(x["stats"]["generation_tps"] for x in runs)
                         ptok = mean(x["stats"]["prompt_tokens"] for x in runs)
                         gtok = mean(x["stats"]["generation_tokens"] for x in runs)
                         peak = mean(
                             x["stats"]["peak_memory_usage"]["inBytes"] for x in runs
                         )
-                        summary = (
+
+                        logger.info(
                             f"prompt_tps={prompt_tps:.2f} gen_tps={gen_tps:.2f}    "
                             f"prompt_tokens={ptok} gen_tokens={gtok}    "
-                            f"peak_memory={format_peak_memory(peak)}"
+                            f"peak_memory={format_peak_memory(peak)}\n"
                         )
-                        if sampler and inference_windows:
-                            joules = sum(
-                                sampler.energy_between(t0, t1)
-                                for t0, t1 in inference_windows
-                            )
-                            inf_seconds = sum(t1 - t0 for t0, t1 in inference_windows)
-                            avg_watts = joules / inf_seconds if inf_seconds > 0 else 0
-                            summary += f"    energy={joules:.1f}J ({avg_watts:.1f}W avg over {inf_seconds:.1f}s inference)"
-                        logger.info(f"{summary}\n")
                     time.sleep(2)
         finally:
-            if sampler:
-                sampler.stop()
-                placement_label = f"{sharding}/{instance_meta}/{n_nodes} nodes"
-                sampler.print_summary(placement_label)
-                placement_metrics = sampler.summarize()
-                if placement_metrics:
-                    all_system_metrics.update(placement_metrics)
-
             try:
                 client.request_json("DELETE", f"/instance/{instance_id}")
             except ExoHttpError as e:
@@ -671,17 +533,11 @@ def main() -> int:
 
             time.sleep(5)
 
-    output: dict[str, Any] = {"runs": all_rows}
-    if cluster_snapshot:
-        output["cluster"] = cluster_snapshot
-    if all_system_metrics:
-        output["system_metrics"] = all_system_metrics
-
     if args.stdout:
-        json.dump(output, sys.stdout, indent=2, ensure_ascii=False)
+        json.dump(all_rows, sys.stdout, indent=2, ensure_ascii=False)
     elif args.json_out:
         with open(args.json_out, "w", encoding="utf-8") as f:
-            json.dump(output, f, indent=2, ensure_ascii=False)
+            json.dump(all_rows, f, indent=2, ensure_ascii=False)
         logger.debug(f"\nWrote results JSON: {args.json_out}")
 
     return 0

bench/harness.py

@@ -69,35 +69,6 @@ class ExoClient:
     def post_bench_chat_completions(self, payload: dict[str, Any]) -> dict[str, Any]:
         return self.request_json("POST", "/bench/chat/completions", body=payload)
 
-    def get_state_path(self, path: str) -> Any:
-        try:
-            return self.request_json("GET", f"/state/{path}")
-        except ExoHttpError as e:
-            if e.status == 404:
-                return None
-            raise
-
-    def get_instance(self, instance_id: str) -> dict[str, Any] | None:
-        return self.get_state_path(f"instances/{instance_id}")
-
-    def get_runner(self, runner_id: str) -> dict[str, Any] | None:
-        return self.get_state_path(f"runners/{runner_id}")
-
-    def get_node_downloads(self, node_id: str) -> list[dict[str, Any]] | None:
-        return self.get_state_path(f"downloads/{node_id}")
-
-    def get_node_disk(self, node_id: str) -> dict[str, Any] | None:
-        return self.get_state_path(f"nodeDisk/{node_id}")
-
-    def get_node_system(self, node_id: str) -> dict[str, Any] | None:
-        return self.get_state_path(f"nodeSystem/{node_id}")
-
-    def get_node_identities(self) -> dict[str, Any] | None:
-        return self.get_state_path("nodeIdentities")
-
-    def get_topology(self) -> dict[str, Any] | None:
-        return self.get_state_path("topology")
-
 
 def unwrap_instance(instance: dict[str, Any]) -> dict[str, Any]:
     if len(instance) != 1:
@@ -126,11 +97,6 @@ def runner_ids_from_instance(instance: dict[str, Any]) -> list[str]:
     return list(runner_to_shard.keys())
 
 
-def node_ids_from_instance(instance: dict[str, Any]) -> list[str]:
-    inner = unwrap_instance(instance)
-    return list(inner["shardAssignments"]["nodeToRunner"].keys())
-
-
 def runner_ready(runner: dict[str, Any]) -> bool:
     return "RunnerReady" in runner
 
@@ -150,12 +116,13 @@ def wait_for_instance_ready(
 ) -> None:
     start_time = time.time()
     instance_existed = False
-    last_loaded: dict[str, int] = {}
     while time.time() - start_time < timeout:
-        instance = client.get_instance(instance_id)
+        state = client.request_json("GET", "/state")
+        instances = state.get("instances", {})
 
-        if instance is None:
+        if instance_id not in instances:
             if instance_existed:
+                # Instance was deleted after being created - likely due to runner failure
                 raise RuntimeError(
                     f"Instance {instance_id} was deleted (runner may have failed)"
                 )
@@ -163,25 +130,18 @@ def wait_for_instance_ready(
             continue
 
         instance_existed = True
-        rids = runner_ids_from_instance(instance)
+        instance = instances[instance_id]
+        runner_ids = runner_ids_from_instance(instance)
+        runners = state.get("runners", {})
 
-        all_ready = True
-        for rid in rids:
-            runner = client.get_runner(rid) or {}
+        # Check for failed runners first
+        for rid in runner_ids:
+            runner = runners.get(rid, {})
             if runner_failed(runner):
                 error_msg = get_runner_failed_message(runner) or "Unknown error"
                 raise RuntimeError(f"Runner {rid} failed: {error_msg}")
-            if "RunnerLoading" in runner:
-                loading = runner["RunnerLoading"]
-                loaded = loading.get("layersLoaded", 0)
-                total = loading.get("totalLayers", 0)
-                if total > 0 and last_loaded.get(rid) != loaded:
-                    last_loaded[rid] = loaded
-                    logger.debug(f"Runner {rid}: loading layers {loaded}/{total}")
-            if not runner_ready(runner):
-                all_ready = False
 
-        if all_ready:
+        if all(runner_ready(runners.get(rid, {})) for rid in runner_ids):
             return
 
         time.sleep(0.1)
@@ -205,23 +165,6 @@ def wait_for_instance_gone(
     raise TimeoutError(f"Instance {instance_id} did not get deleted within {timeout=}")
 
 
-def capture_cluster_snapshot(client: ExoClient) -> dict[str, Any]:
-    snapshot: dict[str, Any] = {}
-    identities = client.get_node_identities()
-    if identities:
-        snapshot["nodeIdentities"] = identities
-    topology = client.get_topology()
-    if topology:
-        snapshot["topology"] = topology
-    node_memory = client.get_state_path("nodeMemory")
-    if node_memory:
-        snapshot["nodeMemory"] = node_memory
-    node_system = client.get_state_path("nodeSystem")
-    if node_system:
-        snapshot["nodeSystem"] = node_system
-    return snapshot
-
-
 def resolve_model_short_id(
     client: ExoClient, model_arg: str, *, force_download: bool = False
 ) -> tuple[str, str]:
@@ -383,11 +326,16 @@ def run_planning_phase(
     node_ids = list(inner["shardAssignments"]["nodeToRunner"].keys())
     runner_to_shard = inner["shardAssignments"]["runnerToShard"]
 
+    state = client.request_json("GET", "/state")
+    downloads = state.get("downloads", {})
+    node_disk = state.get("nodeDisk", {})
+
     needs_download = False
 
     for node_id in node_ids:
-        node_downloads = client.get_node_downloads(node_id) or []
+        node_downloads = downloads.get(node_id, [])
 
+        # Check if model already downloaded on this node
         already_downloaded = any(
             "DownloadCompleted" in p
             and unwrap_instance(p["DownloadCompleted"]["shardMetadata"])["modelCard"][
@@ -401,7 +349,8 @@ def run_planning_phase(
 
         needs_download = True
 
-        disk_info = client.get_node_disk(node_id) or {}
+        # Wait for disk info if settle_deadline is set
+        disk_info = node_disk.get(node_id, {})
         backoff = _SETTLE_INITIAL_BACKOFF_S
         while not disk_info and settle_deadline and time.monotonic() < settle_deadline:
             remaining = settle_deadline - time.monotonic()
@@ -410,7 +359,9 @@ def run_planning_phase(
             )
             time.sleep(min(backoff, remaining))
             backoff = min(backoff * _SETTLE_BACKOFF_MULTIPLIER, _SETTLE_MAX_BACKOFF_S)
-            disk_info = client.get_node_disk(node_id) or {}
+            state = client.request_json("GET", "/state")
+            node_disk = state.get("nodeDisk", {})
+            disk_info = node_disk.get(node_id, {})
 
         if not disk_info:
             logger.warning(f"No disk info for {node_id}, skipping space check")
@@ -426,6 +377,7 @@ def run_planning_phase(
                 f"have {avail // (1024**3)}GB. Use --danger-delete-downloads to free space."
             )
 
+        # Delete from smallest to largest (skip read-only models from EXO_MODELS_PATH)
         completed = [
             (
                 unwrap_instance(p["DownloadCompleted"]["shardMetadata"])["modelCard"][
@@ -462,22 +414,24 @@ def run_planning_phase(
         )
         logger.info(f"Started download on {node_id}")
 
-    # Wait for downloads (no timeout — poll until complete or failed)
-    while True:
+    # Wait for downloads
+    start = time.time()
+    while time.time() - start < timeout:
+        state = client.request_json("GET", "/state")
+        downloads = state.get("downloads", {})
         all_done = True
         for node_id in node_ids:
-            node_downloads = client.get_node_downloads(node_id) or []
             done = any(
                 "DownloadCompleted" in p
                 and unwrap_instance(p["DownloadCompleted"]["shardMetadata"])[
                     "modelCard"
                 ]["modelId"]
                 == full_model_id
-                for p in node_downloads
+                for p in downloads.get(node_id, [])
             )
             failed = [
                 p["DownloadFailed"]["errorMessage"]
-                for p in node_downloads
+                for p in downloads.get(node_id, [])
                 if "DownloadFailed" in p
                 and unwrap_instance(p["DownloadFailed"]["shardMetadata"])["modelCard"][
                     "modelId"
@@ -488,32 +442,13 @@ def run_planning_phase(
                 raise RuntimeError(f"Download failed on {node_id}: {failed[0]}")
             if not done:
                 all_done = False
-                ongoing = [
-                    p
-                    for p in node_downloads
-                    if "DownloadOngoing" in p
-                    and unwrap_instance(p["DownloadOngoing"]["shardMetadata"])[
-                        "modelCard"
-                    ]["modelId"]
-                    == full_model_id
-                ]
-                if ongoing:
-                    prog = ongoing[0]["DownloadOngoing"]["downloadProgress"]
-                    speed_mb = prog.get("speed", 0) / (1024 * 1024)
-                    eta_s = prog.get("etaMs", 0) / 1000
-                    dl_bytes = prog.get("downloaded", {}).get("inBytes", 0)
-                    total_bytes = prog.get("total", {}).get("inBytes", 0)
-                    pct = (dl_bytes / total_bytes * 100) if total_bytes else 0
-                    logger.info(
-                        f"Downloading on {node_id}: {pct:.1f}% @ {speed_mb:.1f} MB/s, "
-                        f"ETA {eta_s:.0f}s "
-                        f"({prog.get('completedFiles', 0)}/{prog.get('totalFiles', 0)} files)"
-                    )
         if all_done:
             if download_t0 is not None:
                 return time.perf_counter() - download_t0
             return None
-        time.sleep(10)
+        time.sleep(1)
+
+    raise TimeoutError("Downloads did not complete in time")
 
 
 def add_common_instance_args(ap: argparse.ArgumentParser) -> None:
@@ -561,8 +496,8 @@ def add_common_instance_args(ap: argparse.ArgumentParser) -> None:
     ap.add_argument(
         "--settle-timeout",
         type=float,
-        default=7200,
-        help="Max seconds to wait for the cluster to produce valid placements (default: 7200).",
+        default=0,
+        help="Max seconds to wait for the cluster to produce valid placements (0 = try once).",
     )
     ap.add_argument(
         "--danger-delete-downloads",

dashboard/package-lock.json

@@ -11,8 +11,7 @@
 				"highlight.js": "^11.11.1",
 				"katex": "^0.16.27",
 				"marked": "^17.0.1",
-				"mode-watcher": "^1.1.0",
-				"pdfjs-dist": "^5.6.205"
+				"mode-watcher": "^1.1.0"
 			},
 			"devDependencies": {
 				"@sveltejs/adapter-static": "^3.0.10",
@@ -519,256 +518,6 @@
 				"@jridgewell/sourcemap-codec": "^1.4.14"
 			}
 		},
-		"node_modules/@napi-rs/canvas": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas/-/canvas-0.1.97.tgz",
-			"integrity": "sha512-8cFniXvrIEnVwuNSRCW9wirRZbHvrD3JVujdS2P5n5xiJZNZMOZcfOvJ1pb66c7jXMKHHglJEDVJGbm8XWFcXQ==",
-			"license": "MIT",
-			"optional": true,
-			"workspaces": [
-				"e2e/*"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			},
-			"optionalDependencies": {
-				"@napi-rs/canvas-android-arm64": "0.1.97",
-				"@napi-rs/canvas-darwin-arm64": "0.1.97",
-				"@napi-rs/canvas-darwin-x64": "0.1.97",
-				"@napi-rs/canvas-linux-arm-gnueabihf": "0.1.97",
-				"@napi-rs/canvas-linux-arm64-gnu": "0.1.97",
-				"@napi-rs/canvas-linux-arm64-musl": "0.1.97",
-				"@napi-rs/canvas-linux-riscv64-gnu": "0.1.97",
-				"@napi-rs/canvas-linux-x64-gnu": "0.1.97",
-				"@napi-rs/canvas-linux-x64-musl": "0.1.97",
-				"@napi-rs/canvas-win32-arm64-msvc": "0.1.97",
-				"@napi-rs/canvas-win32-x64-msvc": "0.1.97"
-			}
-		},
-		"node_modules/@napi-rs/canvas-android-arm64": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-android-arm64/-/canvas-android-arm64-0.1.97.tgz",
-			"integrity": "sha512-V1c/WVw+NzH8vk7ZK/O8/nyBSCQimU8sfMsB/9qeSvdkGKNU7+mxy/bIF0gTgeBFmHpj30S4E9WHMSrxXGQuVQ==",
-			"cpu": [
-				"arm64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"android"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-darwin-arm64": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-darwin-arm64/-/canvas-darwin-arm64-0.1.97.tgz",
-			"integrity": "sha512-ok+SCEF4YejcxuJ9Rm+WWunHHpf2HmiPxfz6z1a/NFQECGXtsY7A4B8XocK1LmT1D7P174MzwPF9Wy3AUAwEPw==",
-			"cpu": [
-				"arm64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"darwin"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-darwin-x64": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-darwin-x64/-/canvas-darwin-x64-0.1.97.tgz",
-			"integrity": "sha512-PUP6e6/UGlclUvAQNnuXCcnkpdUou6VYZfQOQxExLp86epOylmiwLkqXIvpFmjoTEDmPmXrI+coL/9EFU1gKPA==",
-			"cpu": [
-				"x64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"darwin"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-linux-arm-gnueabihf": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-linux-arm-gnueabihf/-/canvas-linux-arm-gnueabihf-0.1.97.tgz",
-			"integrity": "sha512-XyXH2L/cic8eTNtbrXCcvqHtMX/nEOxN18+7rMrAM2XtLYC/EB5s0wnO1FsLMWmK+04ZSLN9FBGipo7kpIkcOw==",
-			"cpu": [
-				"arm"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"linux"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-linux-arm64-gnu": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-linux-arm64-gnu/-/canvas-linux-arm64-gnu-0.1.97.tgz",
-			"integrity": "sha512-Kuq/M3djq0K8ktgz6nPlK7Ne5d4uWeDxPpyKWOjWDK2RIOhHVtLtyLiJw2fuldw7Vn4mhw05EZXCEr4Q76rs9w==",
-			"cpu": [
-				"arm64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"linux"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-linux-arm64-musl": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-linux-arm64-musl/-/canvas-linux-arm64-musl-0.1.97.tgz",
-			"integrity": "sha512-kKmSkQVnWeqg7qdsiXvYxKhAFuHz3tkBjW/zyQv5YKUPhotpaVhpBGv5LqCngzyuRV85SXoe+OFj+Tv0a0QXkQ==",
-			"cpu": [
-				"arm64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"linux"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-linux-riscv64-gnu": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-linux-riscv64-gnu/-/canvas-linux-riscv64-gnu-0.1.97.tgz",
-			"integrity": "sha512-Jc7I3A51jnEOIAXeLsN/M/+Z28LUeakcsXs07FLq9prXc0eYOtVwsDEv913Gr+06IRo34gJJVgT0TXvmz+N2VA==",
-			"cpu": [
-				"riscv64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"linux"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-linux-x64-gnu": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-linux-x64-gnu/-/canvas-linux-x64-gnu-0.1.97.tgz",
-			"integrity": "sha512-iDUBe7AilfuBSRbSa8/IGX38Mf+iCSBqoVKLSQ5XaY2JLOaqz1TVyPFEyIck7wT6mRQhQt5sN6ogfjIDfi74tg==",
-			"cpu": [
-				"x64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"linux"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-linux-x64-musl": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-linux-x64-musl/-/canvas-linux-x64-musl-0.1.97.tgz",
-			"integrity": "sha512-AKLFd/v0Z5fvgqBDqhvqtAdx+fHMJ5t9JcUNKq4FIZ5WH+iegGm8HPdj00NFlCSnm83Fp3Ln8I2f7uq1aIiWaA==",
-			"cpu": [
-				"x64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"linux"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-win32-arm64-msvc": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-win32-arm64-msvc/-/canvas-win32-arm64-msvc-0.1.97.tgz",
-			"integrity": "sha512-u883Yr6A6fO7Vpsy9YE4FVCIxzzo5sO+7pIUjjoDLjS3vQaNMkVzx5bdIpEL+ob+gU88WDK4VcxYMZ6nmnoX9A==",
-			"cpu": [
-				"arm64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"win32"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
-		"node_modules/@napi-rs/canvas-win32-x64-msvc": {
-			"version": "0.1.97",
-			"resolved": "https://registry.npmjs.org/@napi-rs/canvas-win32-x64-msvc/-/canvas-win32-x64-msvc-0.1.97.tgz",
-			"integrity": "sha512-sWtD2EE3fV0IzN+iiQUqr/Q1SwqWhs2O1FKItFlxtdDkikpEj5g7DKQpY3x55H/MAOnL8iomnlk3mcEeGiUMoQ==",
-			"cpu": [
-				"x64"
-			],
-			"license": "MIT",
-			"optional": true,
-			"os": [
-				"win32"
-			],
-			"engines": {
-				"node": ">= 10"
-			},
-			"funding": {
-				"type": "github",
-				"url": "https://github.com/sponsors/Brooooooklyn"
-			}
-		},
 		"node_modules/@polka/url": {
 			"version": "1.0.0-next.29",
 			"resolved": "https://registry.npmjs.org/@polka/url/-/url-1.0.0-next.29.tgz",
@@ -2886,26 +2635,6 @@
 				"node": "^10 || ^12 || ^13.7 || ^14 || >=15.0.1"
 			}
 		},
-		"node_modules/node-readable-to-web-readable-stream": {
-			"version": "0.4.2",
-			"resolved": "https://registry.npmjs.org/node-readable-to-web-readable-stream/-/node-readable-to-web-readable-stream-0.4.2.tgz",
-			"integrity": "sha512-/cMZNI34v//jUTrI+UIo4ieHAB5EZRY/+7OmXZgBxaWBMcW2tGdceIw06RFxWxrKZ5Jp3sI2i5TsRo+CBhtVLQ==",
-			"license": "MIT",
-			"optional": true
-		},
-		"node_modules/pdfjs-dist": {
-			"version": "5.6.205",
-			"resolved": "https://registry.npmjs.org/pdfjs-dist/-/pdfjs-dist-5.6.205.tgz",
-			"integrity": "sha512-tlUj+2IDa7G1SbvBNN74UHRLJybZDWYom+k6p5KIZl7huBvsA4APi6mKL+zCxd3tLjN5hOOEE9Tv7VdzO88pfg==",
-			"license": "Apache-2.0",
-			"engines": {
-				"node": ">=20.19.0 || >=22.13.0 || >=24"
-			},
-			"optionalDependencies": {
-				"@napi-rs/canvas": "^0.1.96",
-				"node-readable-to-web-readable-stream": "^0.4.2"
-			}
-		},
 		"node_modules/picocolors": {
 			"version": "1.1.1",
 			"resolved": "https://registry.npmjs.org/picocolors/-/picocolors-1.1.1.tgz",

dashboard/package.json

@@ -31,7 +31,6 @@
 		"highlight.js": "^11.11.1",
 		"katex": "^0.16.27",
 		"marked": "^17.0.1",
-		"mode-watcher": "^1.1.0",
-		"pdfjs-dist": "^5.6.205"
+		"mode-watcher": "^1.1.0"
 	}
 }

dashboard/src/lib/components/ChatForm.svelte

@@ -1,6 +1,9 @@
 <script lang="ts">
   import {
     isLoading,
+    sendMessage,
+    generateImage,
+    editImage,
     editingImage,
     clearEditingImage,
     selectedChatModel,
@@ -25,7 +28,7 @@
     modelTasks?: Record<string, string[]>;
     modelCapabilities?: Record<string, string[]>;
     onSend?: () => void;
-    onAutoSend: (
+    onAutoSend?: (
       content: string,
       files?: {
         id: string;
@@ -213,10 +216,49 @@
     uploadedFiles = [];
     resetTextareaHeight();
 
-    // Parent controls all send logic (including image routing,
-    // launching non-running models before sending, etc.)
-    onAutoSend(content, files);
+    // When onAutoSend is provided, the parent controls all send logic
+    // (including launching non-running models before sending)
+    if (onAutoSend) {
+      onAutoSend(content, files);
+      onSend?.();
+      setTimeout(() => textareaRef?.focus(), 10);
+      return;
+    }
+
+    // Use image editing if in edit mode
+    if (isEditMode && currentEditingImage && content) {
+      editImage(content, currentEditingImage.imageDataUrl);
+    }
+    // If user attached an image with an ImageToImage model, use edit endpoint
+    else if (
+      currentModel &&
+      modelSupportsImageEditing(currentModel) &&
+      files.length > 0 &&
+      content
+    ) {
+      // Use the first attached image for editing
+      const imageFile = files[0];
+      if (imageFile.preview) {
+        editImage(content, imageFile.preview);
+      }
+    } else if (
+      currentModel &&
+      modelSupportsTextToImage(currentModel) &&
+      content
+    ) {
+      // Use image generation for text-to-image models
+      generateImage(content);
+    } else {
+      sendMessage(
+        content,
+        files,
+        modelSupportsThinking() ? thinkingEnabled : null,
+      );
+    }
+
     onSend?.();
+
+    // Refocus the textarea after sending
     setTimeout(() => textareaRef?.focus(), 10);
   }
 

dashboard/src/lib/components/ChatMessages.svelte

@@ -139,8 +139,6 @@
         return "🖼";
       case "text":
         return "📄";
-      case "pdf":
-        return "📑";
       default:
         return "📎";
     }

dashboard/src/lib/components/FamilyLogos.svelte

@@ -82,18 +82,6 @@
       d="M12.025 1.13c-5.77 0-10.449 4.647-10.449 10.378 0 1.112.178 2.181.503 3.185.064-.222.203-.444.416-.577a.96.96 0 0 1 .524-.15c.293 0 .584.124.84.284.278.173.48.408.71.694.226.282.458.611.684.951v-.014c.017-.324.106-.622.264-.874s.403-.487.762-.543c.3-.047.596.06.787.203s.31.313.4.467c.15.257.212.468.233.542.01.026.653 1.552 1.657 2.54.616.605 1.01 1.223 1.082 1.912.055.537-.096 1.059-.38 1.572.637.121 1.294.187 1.967.187.657 0 1.298-.063 1.921-.178-.287-.517-.44-1.041-.384-1.581.07-.69.465-1.307 1.081-1.913 1.004-.987 1.647-2.513 1.657-2.539.021-.074.083-.285.233-.542.09-.154.208-.323.4-.467a1.08 1.08 0 0 1 .787-.203c.359.056.604.29.762.543s.247.55.265.874v.015c.225-.34.457-.67.683-.952.23-.286.432-.52.71-.694.257-.16.547-.284.84-.285a.97.97 0 0 1 .524.151c.228.143.373.388.43.625l.006.04a10.3 10.3 0 0 0 .534-3.273c0-5.731-4.678-10.378-10.449-10.378M8.327 6.583a1.5 1.5 0 0 1 .713.174 1.487 1.487 0 0 1 .617 2.013c-.183.343-.762-.214-1.102-.094-.38.134-.532.914-.917.71a1.487 1.487 0 0 1 .69-2.803m7.486 0a1.487 1.487 0 0 1 .689 2.803c-.385.204-.536-.576-.916-.71-.34-.12-.92.437-1.103.094a1.487 1.487 0 0 1 .617-2.013 1.5 1.5 0 0 1 .713-.174m-10.68 1.55a.96.96 0 1 1 0 1.921.96.96 0 0 1 0-1.92m13.838 0a.96.96 0 1 1 0 1.92.96.96 0 0 1 0-1.92M8.489 11.458c.588.01 1.965 1.157 3.572 1.164 1.607-.007 2.984-1.155 3.572-1.164.196-.003.305.12.305.454 0 .886-.424 2.328-1.563 3.202-.22-.756-1.396-1.366-1.63-1.32q-.011.001-.02.006l-.044.026-.01.008-.03.024q-.018.017-.035.036l-.032.04a1 1 0 0 0-.058.09l-.014.025q-.049.088-.11.19a1 1 0 0 1-.083.116 1.2 1.2 0 0 1-.173.18q-.035.029-.075.058a1.3 1.3 0 0 1-.251-.243 1 1 0 0 1-.076-.107c-.124-.193-.177-.363-.337-.444-.034-.016-.104-.008-.2.022q-.094.03-.216.087-.06.028-.125.063l-.13.074q-.067.04-.136.086a3 3 0 0 0-.135.096 3 3 0 0 0-.26.219 2 2 0 0 0-.12.121 2 2 0 0 0-.106.128l-.002.002a2 2 0 0 0-.09.132l-.001.001a1.2 1.2 0 0 0-.105.212q-.013.036-.024.073c-1.139-.875-1.563-2.317-1.563-3.203 0-.334.109-.457.305-.454m.836 10.354c.824-1.19.766-2.082-.365-3.194-1.13-1.112-1.789-2.738-1.789-2.738s-.246-.945-.806-.858-.97 1.499.202 2.362c1.173.864-.233 1.45-.685.64-.45-.812-1.683-2.896-2.322-3.295s-1.089-.175-.938.647 2.822 2.813 2.562 3.244-1.176-.506-1.176-.506-2.866-2.567-3.49-1.898.473 1.23 2.037 2.16c1.564.932 1.686 1.178 1.464 1.53s-3.675-2.511-4-1.297c-.323 1.214 3.524 1.567 3.287 2.405-.238.839-2.71-1.587-3.216-.642-.506.946 3.49 2.056 3.522 2.064 1.29.33 4.568 1.028 5.713-.624m5.349 0c-.824-1.19-.766-2.082.365-3.194 1.13-1.112 1.789-2.738 1.789-2.738s.246-.945.806-.858.97 1.499-.202 2.362c-1.173.864.233 1.45.685.64.451-.812 1.683-2.896 2.322-3.295s1.089-.175.938.647-2.822 2.813-2.562 3.244 1.176-.506 1.176-.506 2.866-2.567 3.49-1.898-.473 1.23-2.037 2.16c-1.564.932-1.686 1.178-1.464 1.53s3.675-2.511 4-1.297c.323 1.214-3.524 1.567-3.287 2.405.238.839 2.71-1.587 3.216-.642.506.946-3.49 2.056-3.522 2.064-1.29.33-4.568 1.028-5.713-.624"
     />
   </svg>
-{:else if family === "step"}
-  <svg class="w-6 h-6 {className}" viewBox="0 0 24 24" fill="currentColor">
-    <path
-      d="M22.012 0h1.032v.927H24v.968h-.956V3.78h-1.032V1.896h-1.878v-.97h1.878V0zM2.6 12.371V1.87h.969v10.502h-.97zm10.423.66h10.95v.918h-6.208v9.579h-4.742V13.03zM5.629 3.333v12.356H0v4.51h10.386V8L20.859 8l-.003-4.668-15.227.001z"
-    />
-  </svg>
-{:else if family === "nemotron"}
-  <svg class="w-6 h-6 {className}" viewBox="0 0 24 24" fill="currentColor">
-    <path
-      d="M8.948 8.798v-1.43a6.7 6.7 0 0 1 .424-.018c3.922-.124 6.493 3.374 6.493 3.374s-2.774 3.851-5.75 3.851c-.398 0-.787-.062-1.158-.185v-4.346c1.528.185 1.837.857 2.747 2.385l2.04-1.714s-1.492-1.952-4-1.952a6.016 6.016 0 0 0-.796.035m0-4.735v2.138l.424-.027c5.45-.185 9.01 4.47 9.01 4.47s-4.08 4.964-8.33 4.964c-.37 0-.733-.035-1.095-.097v1.325c.3.035.61.062.91.062 3.957 0 6.82-2.023 9.593-4.408.459.371 2.34 1.263 2.73 1.652-2.633 2.208-8.772 3.984-12.253 3.984-.335 0-.653-.018-.971-.053v1.864H24V4.063zm0 10.326v1.131c-3.657-.654-4.673-4.46-4.673-4.46s1.758-1.944 4.673-2.262v1.237H8.94c-1.528-.186-2.73 1.245-2.73 1.245s.68 2.412 2.739 3.11M2.456 10.9s2.164-3.197 6.5-3.533V6.201C4.153 6.59 0 10.653 0 10.653s2.35 6.802 8.948 7.42v-1.237c-4.84-.6-6.492-5.936-6.492-5.936z"
-    />
-  </svg>
 {:else}
   <svg class="w-6 h-6 {className}" viewBox="0 0 24 24" fill="currentColor">
     <path

dashboard/src/lib/components/FamilySidebar.svelte

@@ -31,7 +31,6 @@
     kimi: "Kimi",
     flux: "FLUX",
     "qwen-image": "Qwen Img",
-    nemotron: "NVIDIA",
   };
 
   function getFamilyName(family: string): string {
@@ -42,20 +41,31 @@
 </script>
 
 <div
-  class="flex flex-col gap-1 py-2 px-1 border-r border-exo-yellow/10 bg-exo-medium-gray/30 min-w-[80px] sm:min-w-[72px] overflow-y-auto scrollbar-hide"
+  class="flex flex-col gap-1 py-2 px-1 border-r border-exo-yellow/10 bg-exo-medium-gray/30 min-w-[72px] sm:min-w-[64px] overflow-y-auto scrollbar-hide"
 >
   <!-- All models (no filter) -->
   <button
     type="button"
     onclick={() => onSelect(null)}
-    class="group flex items-center justify-center px-3 py-2.5 rounded transition-all duration-200 cursor-pointer min-h-[44px] sm:min-h-0 {selectedFamily ===
+    class="group flex flex-col items-center justify-center p-2 sm:p-2 rounded transition-all duration-200 cursor-pointer min-h-[44px] sm:min-h-0 {selectedFamily ===
     null
       ? 'bg-exo-yellow/20 border-l-2 border-exo-yellow'
       : 'hover:bg-white/5 border-l-2 border-transparent'}"
     title="All models"
   >
+    <svg
+      class="w-5 h-5 {selectedFamily === null
+        ? 'text-exo-yellow'
+        : 'text-white/50 group-hover:text-white/70'}"
+      viewBox="0 0 24 24"
+      fill="currentColor"
+    >
+      <path
+        d="M4 8h4V4H4v4zm6 12h4v-4h-4v4zm-6 0h4v-4H4v4zm0-6h4v-4H4v4zm6 0h4v-4h-4v4zm6-10v4h4V4h-4zm-6 4h4V4h-4v4zm6 6h4v-4h-4v4zm0 6h4v-4h-4v4z"
+      />
+    </svg>
     <span
-      class="text-[12px] font-mono font-medium {selectedFamily === null
+      class="text-[9px] font-mono mt-0.5 {selectedFamily === null
         ? 'text-exo-yellow'
         : 'text-white/40 group-hover:text-white/60'}">All</span
     >
@@ -79,7 +89,7 @@
           : "text-white/50 group-hover:text-amber-400/70"}
       />
       <span
-        class="text-[11px] font-mono mt-0.5 {selectedFamily === 'favorites'
+        class="text-[9px] font-mono mt-0.5 {selectedFamily === 'favorites'
           ? 'text-amber-400'
           : 'text-white/40 group-hover:text-white/60'}">Faves</span
       >
@@ -104,7 +114,7 @@
           : "text-white/50 group-hover:text-white/70"}
       />
       <span
-        class="text-[11px] font-mono mt-0.5 {selectedFamily === 'recents'
+        class="text-[9px] font-mono mt-0.5 {selectedFamily === 'recents'
           ? 'text-exo-yellow'
           : 'text-white/40 group-hover:text-white/60'}">Recent</span
       >
@@ -128,7 +138,7 @@
         : "text-white/50 group-hover:text-orange-400/70"}
     />
     <span
-      class="text-[11px] font-mono mt-0.5 {selectedFamily === 'huggingface'
+      class="text-[9px] font-mono mt-0.5 {selectedFamily === 'huggingface'
         ? 'text-orange-400'
         : 'text-white/40 group-hover:text-white/60'}">Hub</span
     >
@@ -154,7 +164,7 @@
           : "text-white/50 group-hover:text-white/70"}
       />
       <span
-        class="text-[11px] font-mono mt-0.5 truncate max-w-full {selectedFamily ===
+        class="text-[9px] font-mono mt-0.5 truncate max-w-full {selectedFamily ===
         family
           ? 'text-exo-yellow'
           : 'text-white/40 group-hover:text-white/60'}"

dashboard/src/lib/components/HeaderNav.svelte

@@ -1,38 +1,21 @@
 <script lang="ts">
   import { browser } from "$app/environment";
 
-  interface Props {
-    showHome?: boolean;
-    onHome?: (() => void) | null;
-    showSidebarToggle?: boolean;
-    sidebarVisible?: boolean;
-    onToggleSidebar?: (() => void) | null;
-    showMobileMenuToggle?: boolean;
-    mobileMenuOpen?: boolean;
-    onToggleMobileMenu?: (() => void) | null;
-    showMobileRightToggle?: boolean;
-    mobileRightOpen?: boolean;
-    onToggleMobileRight?: (() => void) | null;
-    downloadProgress?: {
-      count: number;
-      percentage: number;
-    } | null;
-  }
-
-  let {
-    showHome = true,
-    onHome = null,
-    showSidebarToggle = false,
-    sidebarVisible = true,
-    onToggleSidebar = null,
-    showMobileMenuToggle = false,
-    mobileMenuOpen = false,
-    onToggleMobileMenu = null,
-    showMobileRightToggle = false,
-    mobileRightOpen = false,
-    onToggleMobileRight = null,
-    downloadProgress = null,
-  }: Props = $props();
+  export let showHome = true;
+  export let onHome: (() => void) | null = null;
+  export let showSidebarToggle = false;
+  export let sidebarVisible = true;
+  export let onToggleSidebar: (() => void) | null = null;
+  export let showMobileMenuToggle = false;
+  export let mobileMenuOpen = false;
+  export let onToggleMobileMenu: (() => void) | null = null;
+  export let showMobileRightToggle = false;
+  export let mobileRightOpen = false;
+  export let onToggleMobileRight: (() => void) | null = null;
+  export let downloadProgress: {
+    count: number;
+    percentage: number;
+  } | null = null;
 
   function handleHome(): void {
     if (onHome) {
@@ -276,26 +259,5 @@
       {/if}
       <span class="hidden sm:inline">Downloads</span>
     </a>
-    <a
-      href="/#/integrations"
-      class="text-xs md:text-sm text-white/70 hover:text-exo-yellow transition-colors tracking-wider uppercase flex items-center gap-1.5 md:gap-2 cursor-pointer"
-      title="Integration configs for external tools"
-    >
-      <svg
-        class="w-4 h-4"
-        viewBox="0 0 24 24"
-        fill="none"
-        stroke="currentColor"
-        stroke-width="2"
-        stroke-linecap="round"
-        stroke-linejoin="round"
-      >
-        <path d="M10 13a5 5 0 0 0 7.54.54l3-3a5 5 0 0 0-7.07-7.07l-1.72 1.71" />
-        <path
-          d="M14 11a5 5 0 0 0-7.54-.54l-3 3a5 5 0 0 0 7.07 7.07l1.71-1.71"
-        />
-      </svg>
-      <span class="hidden sm:inline">Integrations</span>
-    </a>
   </nav>
 </header>

dashboard/src/lib/components/IntegrationCard.svelte

@@ -1,52 +0,0 @@
-<script lang="ts">
-  interface Props {
-    title: string;
-    subtitle: string;
-    config: string;
-    description?: string;
-    language?: "json" | "bash";
-  }
-
-  let {
-    title,
-    subtitle,
-    config,
-    description = "",
-    language = "json",
-  }: Props = $props();
-
-  let copied = $state(false);
-
-  async function copyToClipboard() {
-    await navigator.clipboard.writeText(config);
-    copied = true;
-    setTimeout(() => (copied = false), 2000);
-  }
-</script>
-
-<div
-  class="border border-exo-light-gray/20 rounded-lg bg-exo-medium-gray/20 overflow-hidden"
->
-  <div class="flex items-center justify-between px-5 py-4">
-    <div>
-      <h3 class="text-white text-sm font-semibold tracking-wide">{title}</h3>
-      <p class="text-exo-light-gray/60 text-xs mt-0.5 font-mono">{subtitle}</p>
-    </div>
-    <button
-      onclick={copyToClipboard}
-      class="px-3 py-1.5 text-xs rounded border transition-all duration-200 cursor-pointer
-        {copied
-        ? 'border-green-500/50 text-green-400 bg-green-500/10'
-        : 'border-exo-light-gray/30 text-exo-light-gray hover:border-exo-yellow/50 hover:text-exo-yellow'}"
-    >
-      {copied ? "Copied!" : "Copy"}
-    </button>
-  </div>
-  {#if description}
-    <p class="text-exo-light-gray/70 text-xs px-5 pb-3">{description}</p>
-  {/if}
-  <div class="bg-black/30 border-t border-exo-light-gray/10">
-    <pre
-      class="text-xs text-exo-light-gray/90 font-mono p-4 overflow-x-auto whitespace-pre">{config}</pre>
-  </div>
-</div>

dashboard/src/lib/components/ModelCard.svelte

@@ -16,9 +16,7 @@
       perNode?: Array<{
         nodeId: string;
         nodeName: string;
-        status: "completed" | "partial" | "pending" | "downloading";
-        percentage: number;
-        progress: DownloadProgress | null;
+        progress: DownloadProgress;
       }>;
     } | null;
     nodes?: Record<string, NodeInfo>;
@@ -147,7 +145,10 @@
     return `${s}s`;
   }
 
-  const perNode = $derived(downloadStatus?.perNode ?? []);
+  const isDownloading = $derived(downloadStatus?.isDownloading ?? false);
+  const progress = $derived(downloadStatus?.progress);
+  const percentage = $derived(progress?.percentage ?? 0);
+  let expandedNodes = $state<Set<string>>(new Set());
 
   function toggleNodeDetails(nodeId: string): void {
     const next = new Set(expandedNodes);
@@ -586,49 +587,23 @@
       </span>
     </div>
 
-    <!-- Download Status (per-node) -->
-    {#if perNode.length > 0}
+    <!-- Download Status -->
+    {#if isDownloading && progress}
       <div class="mb-2 space-y-1">
-        <div
-          class="text-[10px] font-mono text-white/20 tracking-widest uppercase"
-        >
-          Download progress
+        <div class="flex items-center justify-between text-xs font-mono">
+          <span class="text-blue-400 tracking-wider uppercase">Downloading</span
+          >
+          <span class="text-white/60"
+            >{percentage.toFixed(1)}% &middot; {formatSpeed(progress.speed)}
+            &middot; {formatEta(progress.etaMs)}</span
+          >
+        </div>
+        <div class="h-1 bg-exo-medium-gray/30 rounded overflow-hidden">
+          <div
+            class="h-full bg-blue-500/70 transition-all duration-300"
+            style="width: {percentage}%"
+          ></div>
         </div>
-        {#each perNode as node}
-          <div class="flex items-center gap-2 text-xs font-mono">
-            <span class="text-white/40 w-20 truncate" title={node.nodeId}
-              >{node.nodeName}</span
-            >
-            <div
-              class="flex-1 h-1 bg-exo-medium-gray/30 rounded overflow-hidden"
-            >
-              <div
-                class="h-full transition-all duration-300 {node.status ===
-                'downloading'
-                  ? 'bg-blue-500/70'
-                  : node.status === 'completed'
-                    ? 'bg-exo-yellow/40'
-                    : 'bg-white/20'}"
-                style="width: {node.percentage}%"
-              ></div>
-            </div>
-            <span
-              class="text-right {node.status === 'completed'
-                ? 'text-exo-yellow/60'
-                : node.status === 'downloading'
-                  ? 'text-blue-400/60'
-                  : 'text-white/30'}"
-            >
-              {#if node.status === "downloading" && node.progress}
-                {Math.round(node.percentage)}% {formatSpeed(
-                  node.progress.speed,
-                )}
-              {:else}
-                {node.percentage > 0 ? `${Math.round(node.percentage)}%` : "0%"}
-              {/if}
-            </span>
-          </div>
-        {/each}
       </div>
     {/if}
 
@@ -687,7 +662,15 @@
             {@const allConnections =
               isDebugMode && usedNodes.length > 1
                 ? (() => {
-                    const conns: Array = [];
+                    const conns: Array<{
+                      ip: string;
+                      iface: string | null;
+                      from: string;
+                      to: string;
+                      midX: number;
+                      midY: number;
+                      arrow: string;
+                    }> = [];
                     for (let i = 0; i < usedNodes.length; i++) {
                       for (let j = i + 1; j < usedNodes.length; j++) {
                         const n1 = usedNodes[i];
@@ -699,12 +682,7 @@
                           const toPos = nodePositions[c.to];
                           const arrow =
                             fromPos && toPos ? getArrow(fromPos, toPos) : "→";
-                          conns.push({
-                            ...c,
-                            midX,
-                            midY,
-                            arrow,
-                          });
+                          conns.push({ ...c, midX, midY, arrow });
                         }
                       }
                     }

dashboard/src/lib/components/ModelFilterPopover.svelte

@@ -73,7 +73,7 @@
 
 <!-- svelte-ignore a11y_no_static_element_interactions -->
 <div
-  class="filter-popover absolute right-0 top-full mt-2 w-64 bg-exo-dark-gray border border-exo-yellow/10 rounded-lg shadow-xl z-20"
+  class="filter-popover absolute right-0 top-full mt-2 w-64 bg-exo-dark-gray border border-exo-yellow/10 rounded-lg shadow-xl z-10"
   transition:fly={{ y: -10, duration: 200, easing: cubicOut }}
   onclick={(e) => e.stopPropagation()}
   role="dialog"

dashboard/src/lib/components/ModelPickerModal.svelte

@@ -459,7 +459,6 @@
       "llama",
       "flux",
       "qwen-image",
-      "nemotron",
     ];
     return Array.from(families).sort((a, b) => {
       const aIdx = familyOrder.indexOf(a);

dashboard/src/lib/components/index.ts

@@ -13,4 +13,3 @@ export { default as ModelFilterPopover } from "./ModelFilterPopover.svelte";
 export { default as ModelPickerGroup } from "./ModelPickerGroup.svelte";
 export { default as ModelPickerModal } from "./ModelPickerModal.svelte";
 export { default as ChatModelSelector } from "./ChatModelSelector.svelte";
-export { default as IntegrationCard } from "./IntegrationCard.svelte";

dashboard/src/lib/stores/app.svelte.ts

@@ -257,12 +257,11 @@ interface RawStateResponse {
 }
 
 export interface MessageAttachment {
-  type: "image" | "text" | "file" | "generated-image" | "pdf";
+  type: "image" | "text" | "file" | "generated-image";
   name: string;
   content?: string;
   preview?: string;
   mimeType?: string;
-  pageImages?: string[];
 }
 
 export interface TopLogprob {
@@ -1578,7 +1577,6 @@ class AppStore {
     // Remove messages after user message (including the user message for image requests
     // since generateImage/editImage will re-add it)
     this.messages = this.messages.slice(0, lastUserIndex);
-    this.updateActiveConversation();
 
     switch (requestType) {
       case "image-generation":
@@ -1794,14 +1792,6 @@ class AppStore {
             this.persistConversation(targetConversationId);
           }
         },
-        {
-          generation_stats: (data) => {
-            const stats = data as { generation_tps: number };
-            if (stats.generation_tps > 0) {
-              this.tps = stats.generation_tps;
-            }
-          },
-        },
       );
 
       // Final update
@@ -1999,14 +1989,6 @@ class AppStore {
             this.persistConversation(targetConversationId);
           }
         },
-        {
-          generation_stats: (data) => {
-            const stats = data as { generation_tps: number };
-            if (stats.generation_tps > 0) {
-              this.tps = stats.generation_tps;
-            }
-          },
-        },
       );
 
       // Final cleanup of the message (if conversation still exists)
@@ -2243,7 +2225,6 @@ class AppStore {
       type: string;
       textContent?: string;
       preview?: string;
-      pageImages?: string[];
     }[],
     enableThinking?: boolean | null,
   ): Promise<void> {
@@ -2279,20 +2260,6 @@ class AppStore {
             preview: file.preview,
             mimeType: file.type,
           });
-        } else if (
-          file.pageImages ||
-          (file.textContent && file.type === "application/pdf")
-        ) {
-          attachments.push({
-            type: "pdf",
-            name: file.name,
-            content: file.textContent,
-            pageImages: file.pageImages,
-            mimeType: file.type,
-          });
-          if (file.textContent) {
-            fileContext += `\n\n[File: ${file.name}]\n\`\`\`\n${file.textContent}\n\`\`\``;
-          }
         } else if (file.textContent) {
           attachments.push({
             type: "text",
@@ -2360,70 +2327,13 @@ class AppStore {
       const apiMessages = [
         systemPrompt,
         ...targetConversation.messages.slice(0, -1).map((m) => {
-          // Check if this message has image or PDF attachments
-          const visualAttachments = m.attachments?.filter(
-            (a) =>
-              (a.type === "image" && a.preview) ||
-              (a.type === "pdf" && a.pageImages?.length),
-          );
-
-          if (visualAttachments && visualAttachments.length > 0) {
-            // Build multimodal content array (OpenAI vision format)
-            const contentParts: Array<
-              | { type: "text"; text: string }
-              | { type: "image_url"; image_url: { url: string } }
-            > = [];
-
-            // Add image parts first
-            for (const att of visualAttachments) {
-              if (att.type === "image" && att.preview) {
-                contentParts.push({
-                  type: "image_url",
-                  image_url: { url: att.preview },
-                });
-              } else if (att.type === "pdf" && att.pageImages) {
-                for (const pageImg of att.pageImages) {
-                  contentParts.push({
-                    type: "image_url",
-                    image_url: { url: pageImg },
-                  });
-                }
-              }
-            }
-
-            // Build text content including any text/pdf file attachments
-            let textContent = m.content;
-            if (m.attachments) {
-              for (const attachment of m.attachments) {
-                if (
-                  (attachment.type === "text" || attachment.type === "pdf") &&
-                  attachment.content
-                ) {
-                  textContent += `\n\n[File: ${attachment.name}]\n\`\`\`\n${attachment.content}\n\`\`\``;
-                }
-              }
-            }
-
-            if (textContent) {
-              contentParts.push({ type: "text", text: textContent });
-            }
-
-            return {
-              role: m.role,
-              content: contentParts,
-            };
-          }
-
-          // Text-only message (original path)
+          // Build content including any text file attachments
           let msgContent = m.content;
 
-          // Add text/pdf attachments as context
+          // Add text attachments as context
           if (m.attachments) {
             for (const attachment of m.attachments) {
-              if (
-                (attachment.type === "text" || attachment.type === "pdf") &&
-                attachment.content
-              ) {
+              if (attachment.type === "text" && attachment.content) {
                 msgContent += `\n\n[File: ${attachment.name}]\n\`\`\`\n${attachment.content}\n\`\`\``;
               }
             }
@@ -2486,7 +2396,7 @@ class AppStore {
 
       let streamedContent = "";
       let streamedThinking = "";
-      let serverTpsReceived = false;
+
       interface ChatCompletionChunk {
         choices?: Array<{
           delta?: { content?: string; reasoning_content?: string };
@@ -2551,6 +2461,7 @@ class AppStore {
             tokenCount += 1;
             this.totalTokens = tokenCount;
 
+            // Update real-time TPS during streaming
             if (firstTokenTime !== null && tokenCount > 1) {
               const elapsed = performance.now() - firstTokenTime;
               this.tps = (tokenCount / elapsed) * 1000;
@@ -2601,24 +2512,16 @@ class AppStore {
               startedAt: this.prefillProgress?.startedAt ?? performance.now(),
             };
           },
-          generation_stats: (data) => {
-            const stats = data as { generation_tps: number };
-
-            if (stats.generation_tps > 0) {
-              this.tps = stats.generation_tps;
-              serverTpsReceived = true;
-            }
-          },
         },
       );
 
       // Clear prefill progress after stream ends
       this.prefillProgress = null;
 
-      // Use server-side TPS if available, otherwise fall back to client-side
-      if (!serverTpsReceived && firstTokenTime !== null && tokenCount > 1) {
+      // Calculate final TPS
+      if (firstTokenTime !== null && tokenCount > 1) {
         const totalGenerationTime = performance.now() - firstTokenTime;
-        this.tps = (tokenCount / totalGenerationTime) * 1000;
+        this.tps = (tokenCount / totalGenerationTime) * 1000; // tokens per second
       }
 
       // Final cleanup of the message (if conversation still exists)
@@ -2723,9 +2626,6 @@ class AppStore {
     this.syncActiveMessagesIfNeeded(targetConversationId);
     this.saveConversationsToStorage();
 
-    const abortController = new AbortController();
-    this.currentAbortController = abortController;
-
     try {
       // Determine the model to use
       const model = this.getModelForRequest(modelId);
@@ -2780,7 +2680,6 @@ class AppStore {
           "Content-Type": "application/json",
         },
         body: JSON.stringify(requestBody),
-        signal: abortController.signal,
       });
 
       if (!response.ok) {
@@ -2920,27 +2819,14 @@ class AppStore {
         );
       }
     } catch (error) {
-      if (abortController.signal.aborted) {
-        this.updateConversationMessage(
-          targetConversationId,
-          assistantMessage.id,
-          (msg) => {
-            msg.content = "Cancelled";
-            msg.attachments = [];
-          },
-        );
-        this.syncActiveMessagesIfNeeded(targetConversationId);
-      } else {
-        console.error("Error generating image:", error);
-        this.handleStreamingError(
-          error,
-          targetConversationId,
-          assistantMessage.id,
-          "Failed to generate image",
-        );
-      }
+      console.error("Error generating image:", error);
+      this.handleStreamingError(
+        error,
+        targetConversationId,
+        assistantMessage.id,
+        "Failed to generate image",
+      );
     } finally {
-      this.currentAbortController = null;
       this.isLoading = false;
       this.saveConversationsToStorage();
     }
@@ -3004,9 +2890,6 @@ class AppStore {
     // Clear editing state
     this.editingImage = null;
 
-    const abortController = new AbortController();
-    this.currentAbortController = abortController;
-
     try {
       // Determine the model to use
       const model = this.getModelForRequest(modelId);
@@ -3068,7 +2951,6 @@ class AppStore {
       const apiResponse = await fetch("/v1/images/edits", {
         method: "POST",
         body: formData,
-        signal: abortController.signal,
       });
 
       if (!apiResponse.ok) {
@@ -3169,27 +3051,14 @@ class AppStore {
         );
       }
     } catch (error) {
-      if (abortController.signal.aborted) {
-        this.updateConversationMessage(
-          targetConversationId,
-          assistantMessage.id,
-          (msg) => {
-            msg.content = "Cancelled";
-            msg.attachments = [];
-          },
-        );
-        this.syncActiveMessagesIfNeeded(targetConversationId);
-      } else {
-        console.error("Error editing image:", error);
-        this.handleStreamingError(
-          error,
-          targetConversationId,
-          assistantMessage.id,
-          "Failed to edit image",
-        );
-      }
+      console.error("Error editing image:", error);
+      this.handleStreamingError(
+        error,
+        targetConversationId,
+        assistantMessage.id,
+        "Failed to edit image",
+      );
     } finally {
-      this.currentAbortController = null;
       this.isLoading = false;
       this.saveConversationsToStorage();
     }
@@ -3378,7 +3247,6 @@ export const sendMessage = (
     type: string;
     textContent?: string;
     preview?: string;
-    pageImages?: string[];
   }[],
   enableThinking?: boolean | null,
 ) => appStore.sendMessage(content, files, enableThinking);

dashboard/src/lib/types/files.ts

@@ -2,15 +2,6 @@
  * File attachment types for the chat interface
  */
 
-import { getDocument, GlobalWorkerOptions, version } from "pdfjs-dist";
-import type { DocumentInitParameters } from "pdfjs-dist/types/src/display/api";
-
-GlobalWorkerOptions.workerSrc = `https://cdn.jsdelivr.net/npm/pdfjs-dist@${version}/build/pdf.worker.mjs`;
-
-const PDF_PAGE_SCALE = 2.0;
-const PDF_MAX_PAGES = 20;
-const PDF_MAX_TEXT_CHARS = 100_000;
-
 export interface ChatUploadedFile {
   id: string;
   name: string;
@@ -19,7 +10,6 @@ export interface ChatUploadedFile {
   file: File;
   preview?: string;
   textContent?: string;
-  pageImages?: string[];
 }
 
 export interface ChatAttachment {
@@ -204,58 +194,6 @@ export function readFileAsText(file: File): Promise<string> {
   });
 }
 
-async function extractPdfContent(
-  file: File,
-): Promise<{ text: string; pageImages: string[] }> {
-  const arrayBuffer = await file.arrayBuffer();
-  const pdf = await getDocument({
-    data: new Uint8Array(arrayBuffer),
-    useSystemFonts: true,
-  } as DocumentInitParameters).promise;
-
-  const numPages = Math.min(pdf.numPages, PDF_MAX_PAGES);
-  const pageTexts: string[] = [];
-  const pageImages: string[] = [];
-
-  for (let i = 1; i <= numPages; i++) {
-    const page = await pdf.getPage(i);
-
-    const content = await page.getTextContent();
-    const strings = content.items
-      .filter((item: any) => "str" in item)
-      .map((item: any) => item.str as string);
-    pageTexts.push(strings.join(" "));
-
-    const viewport = page.getViewport({ scale: PDF_PAGE_SCALE });
-    const canvas = new OffscreenCanvas(viewport.width, viewport.height);
-    const ctx = canvas.getContext("2d");
-    if (ctx) {
-      await page.render({ canvasContext: ctx as any, viewport }).promise;
-      const blob = await canvas.convertToBlob({
-        type: "image/jpeg",
-        quality: 0.8,
-      });
-      const reader = new FileReader();
-      const dataUrl = await new Promise<string>((resolve, reject) => {
-        reader.onload = () => resolve(reader.result as string);
-        reader.onerror = () => reject(reader.error);
-        reader.readAsDataURL(blob);
-      });
-      pageImages.push(dataUrl);
-    }
-  }
-
-  let text = pageTexts.join("\n\n").trim();
-  if (text.length > PDF_MAX_TEXT_CHARS) {
-    text = text.slice(0, PDF_MAX_TEXT_CHARS) + "\n\n[truncated]";
-  }
-  if (pdf.numPages > PDF_MAX_PAGES) {
-    text += `\n\n[showing ${PDF_MAX_PAGES} of ${pdf.numPages} pages]`;
-  }
-
-  return { text, pageImages };
-}
-
 /**
  * Process uploaded files into ChatUploadedFile format
  */
@@ -285,12 +223,7 @@ export async function processUploadedFiles(
         const textContent = await readFileAsText(file);
         results.push({ ...base, textContent });
       } else if (category === "pdf") {
-        const { text, pageImages } = await extractPdfContent(file);
-        results.push({
-          ...base,
-          textContent: text || undefined,
-          pageImages: pageImages.length > 0 ? pageImages : undefined,
-        });
+        results.push(base);
       } else if (category === "audio") {
         const preview = await readFileAsDataURL(file);
         results.push({ ...base, preview });

dashboard/src/routes/+page.svelte

@@ -42,10 +42,6 @@
     setSelectedChatModel,
     selectedChatModel,
     sendMessage,
-    thinkingEnabled,
-    generateImage,
-    editImage,
-    editingImage,
     messages,
     debugMode,
     toggleDebugMode,
@@ -265,7 +261,6 @@
 
   let mounted = $state(false);
   let localNodeId = $state<string | null>(null);
-  let pendingFirefoxQuery = $state<string | null>(null); // ?q= param deferred until state loads
 
   // ── Onboarding wizard state ──
   const ONBOARDING_COMPLETE_KEY = "exo-onboarding-complete";
@@ -839,52 +834,6 @@
     if (!model?.tasks) return false;
     return model.tasks.includes("ImageToImage");
   }
-
-  // Route a message to the correct endpoint based on model capabilities.
-  // Image models go to generateImage/editImage; text models go to sendMessage.
-  function routeMessage(
-    content: string,
-    files?: {
-      id: string;
-      name: string;
-      type: string;
-      textContent?: string;
-      preview?: string;
-    }[],
-  ) {
-    const model = selectedChatModel();
-    if (!model) {
-      sendMessage(content, files, thinkingEnabled());
-      return;
-    }
-
-    const currentEditImage = editingImage();
-
-    // Image editing mode (explicit edit or attached image with ImageToImage model)
-    if (currentEditImage && content && modelSupportsImageEditing(model)) {
-      editImage(content, currentEditImage.imageDataUrl);
-      return;
-    }
-    if (
-      modelSupportsImageEditing(model) &&
-      files?.length &&
-      files[0].preview &&
-      content
-    ) {
-      editImage(content, files[0].preview);
-      return;
-    }
-
-    // Text-to-image generation
-    if (modelSupportsImageGeneration(model) && content) {
-      generateImage(content);
-      return;
-    }
-
-    // Default: text chat
-    sendMessage(content, files, thinkingEnabled());
-  }
-
   let selectedSharding = $state<"Pipeline" | "Tensor">("Pipeline");
   type InstanceMeta = "MlxRing" | "MlxJaccl";
 
@@ -1310,20 +1259,6 @@
       return;
     }
 
-    // Firefox AI sidebar integration: handle ?q= query parameter
-    // Firefox's built-in AI sidebar (about:config: browser.ml.chat.enabled) sends
-    // the user's prompt as ?q=<URL-encoded prompt> to the configured provider URL.
-    // See: https://support.mozilla.org/en-US/kb/ai-chatbot
-    const queryParam = params.get("q");
-    if (queryParam) {
-      // Clean up the URL to prevent re-submission on page refresh
-      window.history.replaceState({}, "", window.location.pathname);
-      // Defer the auto-send until cluster state is loaded (topologyData,
-      // instances, availableMemory) so that model auto-selection works
-      // correctly. The $effect below will pick this up once data is ready.
-      pendingFirefoxQuery = queryParam;
-    }
-
     // Check server-side onboarding state (persisted in ~/.exo)
     try {
       const res = await fetch("/onboarding");
@@ -1344,18 +1279,6 @@
     }
   });
 
-  // Deferred Firefox AI sidebar auto-send: wait for cluster state and model
-  // list before submitting. Both data (from /state polling) and models (from
-  // the async /models fetch in onMount) must be loaded for handleAutoSend to
-  // correctly auto-select a model.
-  $effect(() => {
-    if (pendingFirefoxQuery && data && models.length > 0) {
-      const query = pendingFirefoxQuery;
-      pendingFirefoxQuery = null;
-      handleChatSend(query);
-    }
-  });
-
   async function fetchModels() {
     try {
       const response = await fetch("/models");
@@ -1563,44 +1486,34 @@
   }
 
   // Helper to get download status for a model (checks all downloads for matching model ID)
-  type NodeDownloadStatus = {
-    nodeId: string;
-    nodeName: string;
-    status: "completed" | "partial" | "pending" | "downloading";
-    percentage: number;
-    progress: DownloadProgress | null;
-  };
-
-  // Shared helper: collect per-node download status for a model across a set of nodes.
-  // Handles deduplication, entry parsing, and aggregation in one place.
-  function collectDownloadStatus(
-    modelId: string,
-    nodeIds?: string[],
-  ): {
+  function getModelDownloadStatus(modelId: string): {
     isDownloading: boolean;
     progress: DownloadProgress | null;
-    perNode: NodeDownloadStatus[];
-    failedError: string | null;
+    perNode: Array<{
+      nodeId: string;
+      nodeName: string;
+      progress: DownloadProgress;
+    }>;
   } {
-    const empty = {
-      isDownloading: false,
-      progress: null,
-      perNode: [] as NodeDownloadStatus[],
-      failedError: null,
-    };
-
     if (!downloadsData || Object.keys(downloadsData).length === 0) {
-      return empty;
+      return { isDownloading: false, progress: null, perNode: [] };
     }
 
-    // Deduplicate by nodeId — a node can have multiple entries for the same model
-    // (e.g. PipelineShardMetadata + TensorShardMetadata). Keep the last entry,
-    // which is the most recently applied event.
-    const perNodeMap = new Map<string, NodeDownloadStatus>();
+    let totalBytes = 0;
+    let downloadedBytes = 0;
+    let totalSpeed = 0;
+    let completedFiles = 0;
+    let totalFiles = 0;
+    let isDownloading = false;
+    const allFiles: DownloadProgress["files"] = [];
+    const perNode: Array<{
+      nodeId: string;
+      nodeName: string;
+      progress: DownloadProgress;
+    }> = [];
 
-    const nodeIdSet = nodeIds ? new Set(nodeIds) : null;
+    // Check all nodes for downloads matching this model
     for (const [nodeId, nodeDownloads] of Object.entries(downloadsData)) {
-      if (nodeIdSet && !nodeIdSet.has(nodeId)) continue;
       if (!Array.isArray(nodeDownloads)) continue;
 
       for (const downloadWrapped of nodeDownloads) {
@@ -1613,118 +1526,46 @@
         const downloadPayload = (downloadWrapped as Record<string, unknown>)[
           downloadKind
         ] as Record<string, unknown>;
+
+        if (downloadKind !== "DownloadOngoing") continue;
         if (!downloadPayload) continue;
 
         const downloadModelId = extractModelIdFromDownload(downloadPayload);
-        if (!downloadModelId || downloadModelId !== modelId) continue;
-
-        // DownloadFailed — return with any data collected so far
-        if (downloadKind === "DownloadFailed") {
-          return {
-            isDownloading: false,
-            progress: null,
-            perNode: Array.from(perNodeMap.values()),
-            failedError:
-              (downloadPayload.errorMessage as string) ||
-              (downloadPayload.error_message as string) ||
-              "Download failed",
-          };
-        }
 
+        // Match if the model ID contains or equals the requested model
+        // (handles cases like "mlx-community/Meta-Llama..." matching)
         if (
-          downloadKind !== "DownloadOngoing" &&
-          downloadKind !== "DownloadPending" &&
-          downloadKind !== "DownloadCompleted"
-        )
-          continue;
-
-        const nodeName =
-          data?.nodes?.[nodeId]?.friendly_name ?? nodeId.slice(0, 8);
-
-        if (downloadKind === "DownloadCompleted") {
-          perNodeMap.set(nodeId, {
-            nodeId,
-            nodeName,
-            status: "completed",
-            percentage: 100,
-            progress: null,
-          });
-          continue;
+          !downloadModelId ||
+          !downloadModelId.includes(modelId.split("/").pop() || modelId)
+        ) {
+          // Try exact match or partial match
+          if (downloadModelId !== modelId) continue;
         }
 
-        if (downloadKind === "DownloadPending") {
-          const pendingDownloaded = getBytes(
-            downloadPayload.downloaded ??
-              downloadPayload.downloaded_bytes ??
-              downloadPayload.downloadedBytes,
-          );
-          const pendingTotal = getBytes(
-            downloadPayload.total ??
-              downloadPayload.total_bytes ??
-              downloadPayload.totalBytes,
-          );
-          if (pendingDownloaded <= 0 && pendingTotal <= 0) continue;
-          const pct =
-            pendingTotal > 0 ? (pendingDownloaded / pendingTotal) * 100 : 0;
-          perNodeMap.set(nodeId, {
-            nodeId,
-            nodeName,
-            status: pendingDownloaded > 0 ? "partial" : "pending",
-            percentage: pct,
-            progress: null,
-          });
-          continue;
-        }
-
-        // DownloadOngoing
-        const progress = parseDownloadProgress(downloadPayload);
-        if (
-          !progress ||
-          (progress.downloadedBytes <= 0 && progress.totalBytes <= 0)
-        )
-          continue;
-
-        perNodeMap.set(nodeId, {
-          nodeId,
-          nodeName,
-          status: "downloading",
-          percentage: progress.percentage,
-          progress,
-        });
-      }
-    }
-
-    // Aggregate from deduplicated per-node entries
-    const perNode = Array.from(perNodeMap.values());
-    let totalBytes = 0;
-    let downloadedBytes = 0;
-    let totalSpeed = 0;
-    let completedFiles = 0;
-    let totalFiles = 0;
-    let isDownloading = false;
-    const allFiles: DownloadProgress["files"] = [];
-
-    for (const node of perNode) {
-      if (node.status === "downloading" && node.progress) {
         isDownloading = true;
-        totalBytes += node.progress.totalBytes;
-        downloadedBytes += node.progress.downloadedBytes;
-        totalSpeed += node.progress.speed;
-        completedFiles += node.progress.completedFiles;
-        totalFiles += node.progress.totalFiles;
-        allFiles.push(...node.progress.files);
+
+        const progress = parseDownloadProgress(downloadPayload);
+        if (progress) {
+          // Sum all values across nodes - each node downloads independently
+          totalBytes += progress.totalBytes;
+          downloadedBytes += progress.downloadedBytes;
+          totalSpeed += progress.speed;
+          completedFiles += progress.completedFiles;
+          totalFiles += progress.totalFiles;
+          allFiles.push(...progress.files);
+
+          const nodeName =
+            data?.nodes?.[nodeId]?.friendly_name ?? nodeId.slice(0, 8);
+          perNode.push({ nodeId, nodeName, progress });
+        }
       }
     }
 
     if (!isDownloading) {
-      return {
-        isDownloading: false,
-        progress: null,
-        perNode,
-        failedError: null,
-      };
+      return { isDownloading: false, progress: null, perNode: [] };
     }
 
+    // ETA = total remaining bytes / total speed across all nodes
     const remainingBytes = totalBytes - downloadedBytes;
     const etaMs = totalSpeed > 0 ? (remainingBytes / totalSpeed) * 1000 : 0;
 
@@ -1741,21 +1582,9 @@
         files: allFiles,
       },
       perNode,
-      failedError: null,
     };
   }
 
-  function getModelDownloadStatus(
-    modelId: string,
-    nodeIds?: string[],
-  ): {
-    isDownloading: boolean;
-    progress: DownloadProgress | null;
-    perNode: NodeDownloadStatus[];
-  } {
-    return collectDownloadStatus(modelId, nodeIds);
-  }
-
   // Helper to get download status for an instance
   function getInstanceDownloadStatus(
     instanceId: string,
@@ -1766,9 +1595,26 @@
     errorMessage: string | null;
     progress: DownloadProgress | null;
     statusText: string;
-    perNode: NodeDownloadStatus[];
+    perNode: Array<{
+      nodeId: string;
+      nodeName: string;
+      progress: DownloadProgress;
+    }>;
   } {
-    // Unwrap the instance to get shard assignments
+    if (!downloadsData || Object.keys(downloadsData).length === 0) {
+      // No download data yet — defer to runner status instead of assuming RUNNING
+      const statusInfo = deriveInstanceStatus(instanceWrapped);
+      return {
+        isDownloading: false,
+        isFailed: false,
+        errorMessage: null,
+        progress: null,
+        statusText: statusInfo.statusText,
+        perNode: [],
+      };
+    }
+
+    // Unwrap the instance
     const [instanceTag, instance] = getTagged(instanceWrapped);
     if (!instance || typeof instance !== "object") {
       return {
@@ -1788,45 +1634,100 @@
         modelId?: string;
       };
     };
-    const instanceModelId = inst.shardAssignments?.modelId;
-
-    if (!instanceModelId) {
-      const statusInfo = deriveInstanceStatus(instanceWrapped);
-      return {
-        isDownloading: false,
-        isFailed: statusInfo.statusText === "FAILED",
-        errorMessage: null,
-        progress: null,
-        statusText: statusInfo.statusText,
-        perNode: [],
-      };
-    }
-
-    // Get node IDs assigned to this instance
     const nodeToRunner = inst.shardAssignments?.nodeToRunner || {};
     const runnerToShard = inst.shardAssignments?.runnerToShard || {};
+    const instanceModelId = inst.shardAssignments?.modelId;
+
+    // Build reverse mapping: runnerId -> nodeId
     const runnerToNode: Record<string, string> = {};
     for (const [nodeId, runnerId] of Object.entries(nodeToRunner)) {
       runnerToNode[runnerId] = nodeId;
     }
-    const instanceNodeIds = Object.keys(runnerToShard)
-      .map((runnerId) => runnerToNode[runnerId])
-      .filter(Boolean);
 
-    const result = collectDownloadStatus(instanceModelId, instanceNodeIds);
+    let totalBytes = 0;
+    let downloadedBytes = 0;
+    let totalSpeed = 0;
+    let completedFiles = 0;
+    let totalFiles = 0;
+    let isDownloading = false;
+    const allFiles: DownloadProgress["files"] = [];
+    const perNode: Array<{
+      nodeId: string;
+      nodeName: string;
+      progress: DownloadProgress;
+    }> = [];
 
-    if (result.failedError) {
-      return {
-        isDownloading: false,
-        isFailed: true,
-        errorMessage: result.failedError,
-        progress: null,
-        statusText: "FAILED",
-        perNode: [],
-      };
+    // Check downloads for nodes that are part of this instance
+    for (const runnerId of Object.keys(runnerToShard)) {
+      const nodeId = runnerToNode[runnerId];
+      if (!nodeId) continue;
+
+      const nodeDownloads = downloadsData[nodeId];
+      if (!Array.isArray(nodeDownloads)) continue;
+
+      for (const downloadWrapped of nodeDownloads) {
+        if (!downloadWrapped || typeof downloadWrapped !== "object") continue;
+
+        const keys = Object.keys(downloadWrapped as Record<string, unknown>);
+        if (keys.length !== 1) continue;
+
+        const downloadKind = keys[0];
+        const downloadPayload = (downloadWrapped as Record<string, unknown>)[
+          downloadKind
+        ] as Record<string, unknown>;
+
+        // Handle DownloadFailed - return immediately with error info
+        if (downloadKind === "DownloadFailed") {
+          const downloadModelId = extractModelIdFromDownload(downloadPayload);
+          if (
+            instanceModelId &&
+            downloadModelId &&
+            downloadModelId === instanceModelId
+          ) {
+            return {
+              isDownloading: false,
+              isFailed: true,
+              errorMessage:
+                (downloadPayload.errorMessage as string) || "Download failed",
+              progress: null,
+              statusText: "FAILED",
+              perNode: [],
+            };
+          }
+        }
+
+        if (downloadKind !== "DownloadOngoing") continue;
+        if (!downloadPayload) continue;
+
+        // Check if this download is for this instance's model
+        const downloadModelId = extractModelIdFromDownload(downloadPayload);
+        if (
+          instanceModelId &&
+          downloadModelId &&
+          downloadModelId === instanceModelId
+        ) {
+          isDownloading = true;
+
+          const progress = parseDownloadProgress(downloadPayload);
+          if (progress) {
+            // Sum all values across nodes - each node downloads independently
+            totalBytes += progress.totalBytes;
+            downloadedBytes += progress.downloadedBytes;
+            totalSpeed += progress.speed;
+            completedFiles += progress.completedFiles;
+            totalFiles += progress.totalFiles;
+            allFiles.push(...progress.files);
+
+            const nodeName =
+              data?.nodes?.[nodeId]?.friendly_name ?? nodeId.slice(0, 8);
+            perNode.push({ nodeId, nodeName, progress });
+          }
+        }
+      }
     }
 
-    if (!result.isDownloading) {
+    if (!isDownloading) {
+      // Check runner status for other states
       const statusInfo = deriveInstanceStatus(instanceWrapped);
       return {
         isDownloading: false,
@@ -1834,17 +1735,30 @@
         errorMessage: null,
         progress: null,
         statusText: statusInfo.statusText,
-        perNode: result.perNode,
+        perNode: [],
       };
     }
 
+    // ETA = total remaining bytes / total speed across all nodes
+    const remainingBytes = totalBytes - downloadedBytes;
+    const etaMs = totalSpeed > 0 ? (remainingBytes / totalSpeed) * 1000 : 0;
+
     return {
       isDownloading: true,
       isFailed: false,
       errorMessage: null,
-      progress: result.progress,
+      progress: {
+        totalBytes,
+        downloadedBytes,
+        speed: totalSpeed,
+        etaMs,
+        percentage: totalBytes > 0 ? (downloadedBytes / totalBytes) * 100 : 0,
+        completedFiles,
+        totalFiles,
+        files: allFiles,
+      },
       statusText: "DOWNLOADING",
-      perNode: result.perNode,
+      perNode,
     };
   }
 
@@ -2872,7 +2786,7 @@
         // Running model is same or better tier — use it directly
         setSelectedChatModel(bestRunning.id);
         if (!chatStarted) createConversation();
-        routeMessage(content, files);
+        sendMessage(content, files);
         return;
       }
     }
@@ -2889,7 +2803,7 @@
     if (hasRunningInstance(autoModel.id)) {
       setSelectedChatModel(autoModel.id);
       if (!chatStarted) createConversation();
-      routeMessage(content, files);
+      sendMessage(content, files);
       return;
     }
 
@@ -3042,7 +2956,7 @@
       if (pendingAutoMessage) {
         const msg = pendingAutoMessage;
         pendingAutoMessage = null;
-        routeMessage(msg.content, msg.files);
+        sendMessage(msg.content, msg.files);
       }
       return;
     }
@@ -3121,7 +3035,7 @@
     // Model is selected and running — send directly
     if (model && hasRunningInstance(model)) {
       chatLaunchState = "ready";
-      routeMessage(content, files);
+      sendMessage(content, files, null);
       return;
     }
 
@@ -4602,7 +4516,7 @@
                 type="button"
                 onclick={() => {
                   completeOnboarding();
-                  sendMessage(chip, undefined, thinkingEnabled());
+                  sendMessage(chip);
                 }}
                 class="px-4 py-2 rounded-full border border-white/10 bg-white/5 text-sm text-white/60 hover:bg-white/10 hover:text-white/80 hover:border-white/20 transition-all duration-200 cursor-pointer"
               >
@@ -5341,10 +5255,10 @@
                             <div
                               class="mt-2 space-y-2 max-h-48 overflow-y-auto pr-1"
                             >
-                              {#each downloadInfo.perNode.filter((n) => n.status === "downloading" && n.progress) as nodeProg}
+                              {#each downloadInfo.perNode as nodeProg}
                                 {@const nodePercent = Math.min(
                                   100,
-                                  Math.max(0, nodeProg.percentage),
+                                  Math.max(0, nodeProg.progress.percentage),
                                 )}
                                 {@const isExpanded =
                                   instanceDownloadExpandedNodes.has(
@@ -5400,17 +5314,15 @@
                                     >
                                       <span
                                         >{formatBytes(
-                                          nodeProg.progress?.downloadedBytes ??
-                                            0,
+                                          nodeProg.progress.downloadedBytes,
                                         )} / {formatBytes(
-                                          nodeProg.progress?.totalBytes ?? 0,
+                                          nodeProg.progress.totalBytes,
                                         )}</span
                                       >
                                       <span
-                                        >{formatSpeed(
-                                          nodeProg.progress?.speed ?? 0,
-                                        )} • ETA {formatEta(
-                                          nodeProg.progress?.etaMs ?? 0,
+                                        >{formatSpeed(nodeProg.progress.speed)} •
+                                        ETA {formatEta(
+                                          nodeProg.progress.etaMs,
                                         )}</span
                                       >
                                     </div>
@@ -5418,14 +5330,14 @@
 
                                   {#if isExpanded}
                                     <div class="mt-2 space-y-1.5">
-                                      {#if nodeProg.progress?.files ?? [].length === 0}
+                                      {#if nodeProg.progress.files.length === 0}
                                         <div
                                           class="text-[11px] font-mono text-exo-light-gray/70"
                                         >
                                           No file details reported.
                                         </div>
                                       {:else}
-                                        {#each nodeProg.progress?.files ?? [] as f}
+                                        {#each nodeProg.progress.files as f}
                                           {@const filePercent = Math.min(
                                             100,
                                             Math.max(0, f.percentage ?? 0),
@@ -5901,15 +5813,12 @@
                 )}
                 {@const allPreviews = filteredPreviews()}
                 {#if selectedModel && allPreviews.length > 0}
+                  {@const downloadStatus = getModelDownloadStatus(
+                    selectedModel.id,
+                  )}
                   {@const tags = modelTags()[selectedModel.id] || []}
                   <div class="space-y-3">
                     {#each allPreviews as apiPreview, i}
-                      {@const downloadStatus = getModelDownloadStatus(
-                        selectedModel.id,
-                        apiPreview.memory_delta_by_node
-                          ? Object.keys(apiPreview.memory_delta_by_node)
-                          : undefined,
-                      )}
                       <div
                         role="group"
                         onmouseenter={() => {
@@ -6097,7 +6006,7 @@
                           onclick={() => {
                             chatLaunchState = "idle";
                             selectedChatCategory = null;
-                            sendMessage(prompt, undefined, thinkingEnabled());
+                            sendMessage(prompt);
                           }}
                           class="text-left px-3 py-2.5 text-xs text-exo-light-gray hover:text-white font-mono rounded-lg border border-exo-medium-gray/30 hover:border-exo-yellow/30 bg-exo-dark-gray/30 hover:bg-exo-dark-gray/60 transition-all duration-200 cursor-pointer"
                         >
@@ -6480,10 +6389,10 @@
                               <div
                                 class="mt-2 space-y-2 max-h-48 overflow-y-auto pr-1"
                               >
-                                {#each downloadInfo.perNode.filter((n) => n.status === "downloading" && n.progress) as nodeProg}
+                                {#each downloadInfo.perNode as nodeProg}
                                   {@const nodePercent = Math.min(
                                     100,
-                                    Math.max(0, nodeProg.percentage),
+                                    Math.max(0, nodeProg.progress.percentage),
                                   )}
                                   {@const isExpanded =
                                     instanceDownloadExpandedNodes.has(
@@ -6542,17 +6451,16 @@
                                       >
                                         <span
                                           >{formatBytes(
-                                            nodeProg.progress
-                                              ?.downloadedBytes ?? 0,
+                                            nodeProg.progress.downloadedBytes,
                                           )} / {formatBytes(
-                                            nodeProg.progress?.totalBytes ?? 0,
+                                            nodeProg.progress.totalBytes,
                                           )}</span
                                         >
                                         <span
                                           >{formatSpeed(
-                                            nodeProg.progress?.speed ?? 0,
+                                            nodeProg.progress.speed,
                                           )} • ETA {formatEta(
-                                            nodeProg.progress?.etaMs ?? 0,
+                                            nodeProg.progress.etaMs,
                                           )}</span
                                         >
                                       </div>
@@ -6560,14 +6468,14 @@
 
                                     {#if isExpanded}
                                       <div class="mt-2 space-y-1.5">
-                                        {#if nodeProg.progress?.files ?? [].length === 0}
+                                        {#if nodeProg.progress.files.length === 0}
                                           <div
                                             class="text-[11px] font-mono text-exo-light-gray/70"
                                           >
                                             No file details reported.
                                           </div>
                                         {:else}
-                                          {#each nodeProg.progress?.files ?? [] as f}
+                                          {#each nodeProg.progress.files as f}
                                             {@const filePercent = Math.min(
                                               100,
                                               Math.max(0, f.percentage ?? 0),

dashboard/src/routes/integrations/+page.svelte

@@ -1,577 +0,0 @@
-<script lang="ts">
-  import { browser } from "$app/environment";
-  import { fade } from "svelte/transition";
-  import HeaderNav from "$lib/components/HeaderNav.svelte";
-  import IntegrationCard from "$lib/components/IntegrationCard.svelte";
-  import { instances, refreshState } from "$lib/stores/app.svelte";
-  import { onMount } from "svelte";
-
-  const apiUrl = browser
-    ? window.location.origin.replace("localhost", "127.0.0.1")
-    : "http://127.0.0.1:52415";
-
-  const instancesData = $derived(instances());
-
-  let modelCapabilities = $state<Record<string, string[]>>({});
-  let modelContextLengths = $state<Record<string, number>>({});
-
-  const runningModels = $derived.by(() => {
-    const models: string[] = [];
-    for (const [, wrapper] of Object.entries(instancesData)) {
-      if (wrapper && typeof wrapper === "object") {
-        const values = Object.values(wrapper as Record<string, unknown>);
-        if (values.length > 0) {
-          const instance = values[0];
-          if (instance && typeof instance === "object") {
-            const inst = instance as {
-              shardAssignments?: { modelId?: string };
-            };
-            const modelId = inst.shardAssignments?.modelId;
-            if (modelId && !models.includes(modelId)) {
-              models.push(modelId);
-            }
-          }
-        }
-      }
-    }
-    return models;
-  });
-
-  function estimateParamSize(modelId: string): number {
-    const match = modelId.match(/(\d+(?:\.\d+)?)[Bb]/);
-    return match ? parseFloat(match[1]) : 0;
-  }
-
-  const modelsBySize = $derived(
-    [...runningModels].sort(
-      (a, b) => estimateParamSize(b) - estimateParamSize(a),
-    ),
-  );
-
-  const defaultTiers = $derived.by(() => {
-    const n = modelsBySize.length;
-    if (n === 0)
-      return {
-        opus: "your-model-id",
-        sonnet: "your-model-id",
-        haiku: "your-model-id",
-      };
-    if (n === 1)
-      return {
-        opus: modelsBySize[0],
-        sonnet: modelsBySize[0],
-        haiku: modelsBySize[0],
-      };
-    if (n === 2)
-      return {
-        opus: modelsBySize[0],
-        sonnet: modelsBySize[1],
-        haiku: modelsBySize[1],
-      };
-    return {
-      opus: modelsBySize[0],
-      sonnet: modelsBySize[Math.floor(n / 2)],
-      haiku: modelsBySize[n - 1],
-    };
-  });
-
-  let opusModel = $state("");
-  let sonnetModel = $state("");
-  let haikuModel = $state("");
-
-  $effect(() => {
-    opusModel = defaultTiers.opus;
-    sonnetModel = defaultTiers.sonnet;
-    haikuModel = defaultTiers.haiku;
-  });
-
-  let codexModel = $state("");
-  let codexMcpPath = $state("/Users/username");
-  let openClawModel = $state("");
-  $effect(() => {
-    const def = modelsBySize.length > 0 ? modelsBySize[0] : "your-model-id";
-    codexModel = def;
-    openClawModel = def;
-  });
-
-  const claudeShellCommand = $derived(
-    [
-      `ANTHROPIC_BASE_URL=${apiUrl} \\`,
-      `ANTHROPIC_API_KEY=x \\`,
-      `ANTHROPIC_DEFAULT_OPUS_MODEL=${opusModel} \\`,
-      `ANTHROPIC_DEFAULT_SONNET_MODEL=${sonnetModel} \\`,
-      `ANTHROPIC_DEFAULT_HAIKU_MODEL=${haikuModel} \\`,
-      `API_TIMEOUT_MS=3000000 \\`,
-      `CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 \\`,
-      `claude`,
-    ].join("\n"),
-  );
-
-  const claudeSettingsJson = $derived(
-    JSON.stringify(
-      {
-        env: {
-          ANTHROPIC_BASE_URL: apiUrl,
-          ANTHROPIC_API_KEY: "x",
-          ANTHROPIC_DEFAULT_OPUS_MODEL: opusModel,
-          ANTHROPIC_DEFAULT_SONNET_MODEL: sonnetModel,
-          ANTHROPIC_DEFAULT_HAIKU_MODEL: haikuModel,
-          API_TIMEOUT_MS: "3000000",
-          CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC: "1",
-        },
-      },
-      null,
-      2,
-    ),
-  );
-
-  const openCodeConfig = $derived.by(() => {
-    const models: Record<string, Record<string, unknown>> = {};
-    for (const modelId of runningModels) {
-      const caps = modelCapabilities[modelId] || [];
-      const ctxLen = modelContextLengths[modelId] || 0;
-      const entry: Record<string, unknown> = { name: modelId };
-      if (ctxLen > 0) {
-        entry.limit = { context: ctxLen, output: Math.min(ctxLen, 16384) };
-      }
-      if (caps.includes("vision")) {
-        entry.modalities = { input: ["text", "image"], output: ["text"] };
-      }
-      models[modelId] = entry;
-    }
-    if (Object.keys(models).length === 0) {
-      models["your-model-id"] = { name: "your-model-name" };
-    }
-    const firstModel =
-      runningModels.length > 0 ? runningModels[0] : "your-model-id";
-    return JSON.stringify(
-      {
-        $schema: "https://opencode.ai/config.json",
-        provider: {
-          exo: {
-            npm: "@ai-sdk/openai-compatible",
-            name: "exo",
-            options: {
-              baseURL: `${apiUrl}/v1`,
-              apiKey: "x",
-            },
-            models,
-          },
-        },
-        model: `exo/${firstModel}`,
-      },
-      null,
-      2,
-    );
-  });
-
-  const codexShellCommand = $derived(`EXO_API_KEY=x npx @openai/codex`);
-
-  const codexConfig = $derived(
-    [
-      `model = "${codexModel}"`,
-      `model_provider = "exo"`,
-      ``,
-      `[model_providers.exo]`,
-      `name = "exo"`,
-      `base_url = "${apiUrl}/v1"`,
-      `env_key = "EXO_API_KEY"`,
-      ``,
-      `[mcp_servers.filesystem]`,
-      `command = "npx"`,
-      `args = ["-y", "@modelcontextprotocol/server-filesystem", "${codexMcpPath}"]`,
-    ].join("\n"),
-  );
-
-  const openClawConfig = $derived(
-    JSON.stringify(
-      {
-        gateway: { mode: "local" },
-        models: {
-          providers: {
-            exo: {
-              baseUrl: `${apiUrl}/v1`,
-              apiKey: "x",
-              api: "openai-completions",
-              models: [
-                {
-                  id: openClawModel,
-                  name: "exo local",
-                  input: (modelCapabilities[openClawModel] || []).includes(
-                    "vision",
-                  )
-                    ? ["text", "image"]
-                    : ["text"],
-                },
-              ],
-            },
-          },
-        },
-        agents: {
-          defaults: {
-            model: `exo/${openClawModel}`,
-          },
-        },
-      },
-      null,
-      2,
-    ),
-  );
-
-  const ollamaCommand = $derived(
-    `OLLAMA_HOST=${apiUrl}/ollama ollama run ${modelsBySize.length > 0 ? modelsBySize[0] : "your-model-id"}`,
-  );
-
-  const openWebUiCommand = $derived(
-    [
-      `docker run -d -p 3000:8080 \\`,
-      `  -e OLLAMA_BASE_URL=${apiUrl.replace("localhost", "host.docker.internal")}/ollama \\`,
-      `  -v open-webui:/app/backend/data \\`,
-      `  --name open-webui \\`,
-      `  ghcr.io/open-webui/open-webui:main`,
-    ].join("\n"),
-  );
-
-  const n8nDockerCommand = $derived(
-    [
-      `docker run -d -p 5678:5678 \\`,
-      `  -v n8n_data:/home/node/.n8n \\`,
-      `  --name n8n \\`,
-      `  docker.n8n.io/n8nio/n8n`,
-    ].join("\n"),
-  );
-
-  const n8nCredentialSteps = $derived(
-    [
-      `1. Go to Credentials → Add Credential → search "OpenAI API"`,
-      `2. Set API Key to: x`,
-      `3. Set Base URL to: ${apiUrl.replace("127.0.0.1", "host.docker.internal").replace("localhost", "host.docker.internal")}/v1`,
-      `4. Save the credential`,
-    ].join("\n"),
-  );
-
-  const n8nWorkflowSteps = $derived(
-    [
-      `1. Create a new workflow → "Start from Scratch"`,
-      `2. Add an "AI Agent" or "Basic LLM Chain" node`,
-      `3. Inside it, add an "OpenAI Chat Model" sub-node`,
-      `4. Select the OpenAI credential you just created`,
-      `5. Set Model to "From list" and pick your model (e.g. ${modelsBySize.length > 0 ? modelsBySize[0] : "your-model-id"})`,
-      `6. Optionally toggle "Use Responses API", add Built-in Tools, or click "Add Option" for sampling settings`,
-      `7. Connect a "Chat Trigger" node for interactive chat`,
-      `8. On the Chat Trigger, enable "Allow File Uploads" for vision`,
-    ].join("\n"),
-  );
-
-  const firefoxConfig = $derived(
-    [
-      `1. Open about:config in Firefox`,
-      `2. Set browser.ml.chat.enabled to true`,
-      `3. Set browser.ml.chat.hideLocalhost to false`,
-      `4. Set browser.ml.chat.provider to: ${apiUrl}/`,
-    ].join("\n"),
-  );
-
-  const tabs = [
-    "Claude Code",
-    "OpenCode",
-    "Codex",
-    "OpenClaw",
-    "Open WebUI",
-    "n8n",
-    "Firefox",
-  ] as const;
-  type Tab = (typeof tabs)[number];
-  const stored = browser ? localStorage.getItem("exo-integrations-tab") : null;
-  let activeTab = $state<Tab>(
-    stored && tabs.includes(stored as Tab) ? (stored as Tab) : "Claude Code",
-  );
-  $effect(() => {
-    if (browser) localStorage.setItem("exo-integrations-tab", activeTab);
-  });
-
-  const selectClass =
-    "bg-black/30 border border-exo-light-gray/20 rounded px-2 py-1.5 text-white font-mono text-xs focus:border-exo-yellow/50 focus:outline-none appearance-none cursor-pointer";
-
-  onMount(async () => {
-    refreshState();
-    try {
-      const resp = await fetch("/v1/models");
-      const data = (await resp.json()) as {
-        data: { id: string; capabilities: string[]; context_length: number }[];
-      };
-      const caps: Record<string, string[]> = {};
-      const ctxs: Record<string, number> = {};
-      for (const model of data.data) {
-        caps[model.id] = model.capabilities || [];
-        if (model.context_length > 0) ctxs[model.id] = model.context_length;
-      }
-      modelCapabilities = caps;
-      modelContextLengths = ctxs;
-    } catch {
-      /* ignore */
-    }
-  });
-</script>
-
-<div class="min-h-screen bg-exo-dark-gray flex flex-col">
-  <HeaderNav showHome={true} />
-
-  <main
-    class="flex-1 max-w-3xl mx-auto w-full px-4 md:px-6 py-8"
-    in:fade={{ duration: 200 }}
-  >
-    <div class="mb-8">
-      <h1
-        class="text-white text-xl md:text-2xl font-semibold tracking-wide mb-2"
-      >
-        Integrations
-      </h1>
-      <p class="text-exo-light-gray/60 text-sm">
-        Connect external tools to your exo cluster.
-      </p>
-    </div>
-
-    <!-- Status -->
-    <div class="mb-8">
-      <span class="text-exo-light-gray/70 text-xs uppercase tracking-wider"
-        >API Endpoint</span
-      >
-      <span class="text-white font-mono text-sm ml-2">{apiUrl}</span>
-      {#if runningModels.length > 0}
-        <div class="text-exo-light-gray/50 text-xs mt-2">
-          Running model{runningModels.length > 1 ? "s" : ""}:
-          <ul class="mt-1 space-y-0.5 list-none">
-            {#each runningModels as model}
-              <li class="text-exo-yellow font-mono">{model}</li>
-            {/each}
-          </ul>
-        </div>
-      {:else}
-        <p class="text-exo-light-gray/40 text-xs mt-2 italic">
-          No models currently running
-        </p>
-      {/if}
-    </div>
-
-    <!-- API Endpoints -->
-    <div class="mb-8">
-      <div
-        class="flex flex-col sm:flex-row gap-3 text-xs font-mono text-exo-light-gray/70"
-      >
-        <div
-          class="flex-1 bg-black/20 border border-exo-light-gray/10 rounded px-3 py-2"
-        >
-          <span class="text-exo-light-gray/40 text-[10px] uppercase block mb-1"
-            >OpenAI-compatible</span
-          >
-          <span class="text-white/80">{apiUrl}/v1</span>
-        </div>
-        <div
-          class="flex-1 bg-black/20 border border-exo-light-gray/10 rounded px-3 py-2"
-        >
-          <span class="text-exo-light-gray/40 text-[10px] uppercase block mb-1"
-            >Claude-compatible</span
-          >
-          <span class="text-white/80">{apiUrl}</span>
-        </div>
-        <div
-          class="flex-1 bg-black/20 border border-exo-light-gray/10 rounded px-3 py-2"
-        >
-          <span class="text-exo-light-gray/40 text-[10px] uppercase block mb-1"
-            >Ollama-compatible</span
-          >
-          <span class="text-white/80">{apiUrl}/ollama</span>
-        </div>
-      </div>
-    </div>
-
-    <!-- Tabs -->
-    <div
-      class="flex flex-wrap gap-2 mb-6 border-b border-exo-light-gray/10 pb-3"
-    >
-      {#each tabs as tab}
-        <button
-          onclick={() => (activeTab = tab)}
-          class="px-3 py-1.5 text-xs rounded-md transition-all cursor-pointer
-            {activeTab === tab
-            ? 'bg-exo-yellow/15 text-exo-yellow border border-exo-yellow/30'
-            : 'text-exo-light-gray/60 hover:text-white/80 border border-transparent hover:border-exo-light-gray/20'}"
-        >
-          {tab}
-        </button>
-      {/each}
-    </div>
-
-    <!-- Tab Content -->
-    <div class="space-y-4">
-      {#if activeTab === "Claude Code"}
-        {#if runningModels.length > 1}
-          <div class="grid grid-cols-3 gap-3 text-xs">
-            {#each [{ label: "Opus", bind: () => opusModel, set: (v: string) => (opusModel = v) }, { label: "Sonnet", bind: () => sonnetModel, set: (v: string) => (sonnetModel = v) }, { label: "Haiku", bind: () => haikuModel, set: (v: string) => (haikuModel = v) }] as tier}
-              <div>
-                <span
-                  class="text-exo-light-gray/50 text-[10px] uppercase tracking-wider block mb-1"
-                  >{tier.label}</span
-                >
-                <select
-                  value={tier.bind()}
-                  onchange={(e) =>
-                    tier.set((e.target as HTMLSelectElement).value)}
-                  class="w-full {selectClass}"
-                >
-                  {#each runningModels as model}
-                    <option value={model}>{model.split("/").pop()}</option>
-                  {/each}
-                </select>
-              </div>
-            {/each}
-          </div>
-        {/if}
-        <IntegrationCard
-          title="Shell Command"
-          subtitle="Run in terminal"
-          description="Launch Claude Code with exo as the backend. Paste this into your terminal."
-          config={claudeShellCommand}
-          language="bash"
-        />
-        <IntegrationCard
-          title="Settings File"
-          subtitle="~/.claude/settings.json"
-          description="Or add this to your Claude Code settings for persistent configuration."
-          config={claudeSettingsJson}
-        />
-      {:else if activeTab === "OpenCode"}
-        <IntegrationCard
-          title="Config File"
-          subtitle="opencode.json"
-          description="Add this to your project root or ~/.config/opencode/opencode.json for global config. Vision models automatically get image input modality."
-          config={openCodeConfig}
-        />
-      {:else if activeTab === "Codex"}
-        <div class="flex gap-3 text-xs">
-          {#if runningModels.length > 1}
-            <div>
-              <span
-                class="text-exo-light-gray/50 text-[10px] uppercase tracking-wider block mb-1"
-                >Model</span
-              >
-              <select bind:value={codexModel} class={selectClass}>
-                {#each runningModels as model}
-                  <option value={model}>{model.split("/").pop()}</option>
-                {/each}
-              </select>
-            </div>
-          {/if}
-          <div class="flex-1">
-            <span
-              class="text-exo-light-gray/50 text-[10px] uppercase tracking-wider block mb-1"
-              >MCP Filesystem Path</span
-            >
-            <input
-              type="text"
-              bind:value={codexMcpPath}
-              class="w-full bg-black/30 border border-exo-light-gray/20 rounded px-2 py-1.5 text-white font-mono text-xs focus:border-exo-yellow/50 focus:outline-none"
-            />
-          </div>
-        </div>
-        <IntegrationCard
-          title="Config File"
-          subtitle="~/.codex/config.toml"
-          description="Add this to your Codex CLI config so the model and provider persist."
-          config={codexConfig}
-        />
-        <IntegrationCard
-          title="Shell Command"
-          subtitle="Run in terminal"
-          description="Launch Codex with exo as the backend."
-          config={codexShellCommand}
-          language="bash"
-        />
-      {:else if activeTab === "OpenClaw"}
-        {#if runningModels.length > 1}
-          <div class="text-xs">
-            <span
-              class="text-exo-light-gray/50 text-[10px] uppercase tracking-wider block mb-1"
-              >Model</span
-            >
-            <select bind:value={openClawModel} class={selectClass}>
-              {#each runningModels as model}
-                <option value={model}>{model.split("/").pop()}</option>
-              {/each}
-            </select>
-          </div>
-        {/if}
-        <IntegrationCard
-          title="Config File"
-          subtitle="~/.openclaw/openclaw.json"
-          description="Add this to your OpenClaw config. If you haven't installed OpenClaw yet, run: npm install -g openclaw@latest"
-          config={openClawConfig}
-        />
-        <IntegrationCard
-          title="Setup Commands"
-          subtitle="Run in terminal"
-          description="After saving the config, run these commands to fix metadata and start the gateway."
-          config={`openclaw doctor --fix${(modelCapabilities[openClawModel] || []).includes("vision") ? `\nopenclaw models set-image exo/${openClawModel}` : ""}\nopenclaw gateway &\nopenclaw dashboard`}
-          language="bash"
-        />
-      {:else if activeTab === "Open WebUI"}
-        <IntegrationCard
-          title="1. Start Open WebUI"
-          subtitle="Run in terminal"
-          description="Run this to start Open WebUI."
-          config={openWebUiCommand}
-          language="bash"
-        />
-        <IntegrationCard
-          title="2. Open & Select Model"
-          subtitle="http://localhost:3000"
-          description={`Open http://localhost:3000 in your browser. Select the running model from the dropdown at the top: ${runningModels.length > 0 ? runningModels.join(", ") : "no models running"}`}
-          config={"open http://localhost:3000"}
-          language="bash"
-        />
-        <IntegrationCard
-          title="Ollama CLI"
-          subtitle="Run in terminal"
-          description="Or use the Ollama CLI directly."
-          config={ollamaCommand}
-          language="bash"
-        />
-      {:else if activeTab === "n8n"}
-        <IntegrationCard
-          title="1. Start n8n"
-          subtitle="Run in terminal"
-          description="Start n8n with Docker. If you already have n8n running, skip this step."
-          config={n8nDockerCommand}
-          language="bash"
-        />
-        <IntegrationCard
-          title="2. Open n8n"
-          subtitle="http://localhost:5678"
-          description="Open n8n in your browser. If this is your first time, complete the setup and select 'Start from Scratch' when prompted."
-          config={"open http://localhost:5678"}
-          language="bash"
-        />
-        <IntegrationCard
-          title="3. Add OpenAI Credential"
-          subtitle="n8n UI → Credentials"
-          description="Create an OpenAI credential pointing at your exo cluster."
-          config={n8nCredentialSteps}
-        />
-        <IntegrationCard
-          title="4. Build a Workflow"
-          subtitle="n8n UI → Workflows"
-          description="Create a workflow that uses your exo-powered model."
-          config={n8nWorkflowSteps}
-        />
-      {:else if activeTab === "Firefox"}
-        <IntegrationCard
-          title="Firefox AI Chatbot"
-          subtitle="about:config"
-          description="Use the exo dashboard as Firefox's built-in AI chatbot. Requires Firefox 130+."
-          config={firefoxConfig}
-        />
-      {/if}
-    </div>
-  </main>
-</div>

flake.lock

@@ -164,11 +164,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1773870109,
-        "narHash": "sha256-ZoTdqZP03DcdoyxvpFHCAek4bkPUTUPUF3oCCgc3dP4=",
+        "lastModified": 1763662255,
+        "narHash": "sha256-4bocaOyLa3AfiS8KrWjZQYu+IAta05u3gYZzZ6zXbT0=",
         "owner": "pyproject-nix",
         "repo": "build-system-pkgs",
-        "rev": "b6e74f433b02fa4b8a7965ee24680f4867e2926f",
+        "rev": "042904167604c681a090c07eb6967b4dd4dae88c",
         "type": "github"
       },
       "original": {
@@ -184,11 +184,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1774498001,
-        "narHash": "sha256-wTfdyzzrmpuqt4TQQNqilF91v0m5Mh1stNy9h7a/WK4=",
+        "lastModified": 1764134915,
+        "narHash": "sha256-xaKvtPx6YAnA3HQVp5LwyYG1MaN4LLehpQI8xEdBvBY=",
         "owner": "pyproject-nix",
         "repo": "pyproject.nix",
-        "rev": "794afa6eb588b498344f2eaa36ab1ceb7e6b0b09",
+        "rev": "2c8df1383b32e5443c921f61224b198a2282a657",
         "type": "github"
       },
       "original": {
@@ -280,11 +280,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1774490495,
-        "narHash": "sha256-a9WmQWj8fF7BctZGCoyzpUjP6GJw8H+lxl+zxpGnETk=",
+        "lastModified": 1767701098,
+        "narHash": "sha256-CJhKZnWb3gumR9oTRjFvCg/6lYTGbZRU7xtvcyWIRwU=",
         "owner": "pyproject-nix",
         "repo": "uv2nix",
-        "rev": "18ae62fc5e389e3069854a7c66455c22e31708fc",
+        "rev": "9d357f0d2ce6f5f35ec7959d7e704452352eb4da",
         "type": "github"
       },
       "original": {

flake.nix

@@ -72,7 +72,7 @@
       ];
 
       perSystem =
-        { config, self', pkgs, lib, system, ... }:
+        { config, self', inputs', pkgs, lib, system, ... }:
         let
           # Use pinned nixpkgs for swift-format (swift is broken on x86_64-linux in newer nixpkgs)
           pkgsSwift = import inputs.nixpkgs-swift { inherit system; };
@@ -84,17 +84,6 @@
             config.allowUnfreePredicate = pkg: (pkg.pname or "") == "metal-toolchain";
             overlays = [
               (import ./nix/apple-sdk-overlay.nix)
-              (final: prev: {
-                macmon = prev.macmon.overrideAttrs (_: {
-                  version = "git";
-                  src = final.fetchFromGitHub {
-                    owner = "swiftraccoon";
-                    repo = "macmon";
-                    rev = "9154d234f763fbeffdcb4135d0bbbaf80609699b";
-                    hash = "sha256-CwhilKNbs5XL9/tF5DMwyPBlE/hpmjGNTuxQ36sM50M=";
-                  };
-                });
-              })
             ];
           };
           treefmt = {
@@ -128,13 +117,12 @@
               uvLock = builtins.fromTOML (builtins.readFile ./uv.lock);
               mlxPackage = builtins.head (builtins.filter (p: p.name == "mlx" && p.source ? git) uvLock.package);
               uvLockMlxVersion = mlxPackage.version;
-              uvLockMlxRev = builtins.elemAt (builtins.split "#" mlxPackage.source.git) 2;
             in
             {
               metal-toolchain = pkgs.callPackage ./nix/metal-toolchain.nix { };
               mlx = pkgs.callPackage ./nix/mlx.nix {
                 inherit (self'.packages) metal-toolchain;
-                inherit uvLockMlxVersion uvLockMlxRev;
+                inherit uvLockMlxVersion;
               };
               default = self'.packages.exo;
             }

justfile

@@ -1,8 +1,5 @@
 export NIX_CONFIG := "extra-experimental-features = nix-command flakes"
 
-default: lint fmt
-all: lint fmt check
-
 fmt:
     treefmt || nix fmt
 
@@ -34,10 +31,6 @@ build-dashboard:
 package:
     uv run pyinstaller packaging/pyinstaller/exo.spec
 
-build-app: package
-    xcodebuild build -project app/EXO/EXO.xcodeproj -scheme EXO -configuration Debug -derivedDataPath app/EXO/build
-    @echo "\nBuild complete. Run with:\n  open {{justfile_directory()}}/app/EXO/build/Build/Products/Debug/EXO.app"
-
 clean:
     rm -rf **/__pycache__
     rm -rf target/

nix/mlx.nix

@@ -11,7 +11,6 @@
 , fmt
 , python313Packages
 , uvLockMlxVersion
-, uvLockMlxRev
 }:
 
 assert stdenv.isDarwin;
@@ -42,13 +41,15 @@ let
 
   mlx = stdenv.mkDerivation rec {
     pname = "mlx";
-    version = uvLockMlxVersion;
+    version = let v = "0.30.7.dev20260225+257d5692"; in
+      assert v == uvLockMlxVersion || throw "MLX version mismatch: nix/mlx.nix has ${v} but uv.lock has ${uvLockMlxVersion}. Update both the version and hash in nix/mlx.nix.";
+      v;
     pyproject = true;
 
     src = fetchFromGitHub {
       owner = "rltakashige";
       repo = "mlx-jaccl-fix-small-recv";
-      rev = uvLockMlxRev;
+      rev = "257d5692fc7af6bba3b8afaeb63c549b7d1e43d5";
       hash = "sha256-GosFIWxIB48Egb1MqJrR3xhsUsQeWdRk5rV93USY6wQ=";
     };
 

packaging/pyinstaller/exo.spec

@@ -71,9 +71,7 @@ MACMON_PATH = shutil.which("macmon")
 if MACMON_PATH is None:
     raise SystemExit(
         "macmon binary not found in PATH. "
-        "Install the pinned fork used by exo via: "
-        "cargo install --git https://github.com/swiftraccoon/macmon "
-        "--rev 9154d234f763fbeffdcb4135d0bbbaf80609699b macmon --force"
+        "Install it via: brew install macmon"
     )
 
 BINARIES: list[tuple[str, str]] = [
@@ -122,3 +120,4 @@ coll = COLLECT(
     upx_exclude=[],
     name="exo",
 )
+

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "exo"
-version = "0.3.69"
+version = "0.3.68"
 description = "Exo"
 readme = "README.md"
 requires-python = ">=3.13"
@@ -12,7 +12,7 @@ dependencies = [
   "fastapi>=0.116.1",
   "filelock>=3.18.0",
   "rustworkx>=0.17.1",
-  "huggingface-hub>=1.8.0",
+  "huggingface-hub>=0.33.4",
   "psutil>=7.0.0",
   "loguru>=0.7.3",
   "exo_pyo3_bindings",                         # rust bindings
@@ -25,12 +25,11 @@ dependencies = [
   "openai-harmony>=0.0.8",
   "httpx>=0.28.1",
   "tomlkit>=0.14.0",
-  "mflux==0.17.2",
+  "pillow>=11.0,<12.0",                        # compatibility with mflux
+  "mflux==0.15.5",
   "python-multipart>=0.0.21",
   "msgspec>=0.19.0",
   "zstandard>=0.23.0",
-  "mlx-vlm>=0.3.11",
-  "transformers>=5.0.0,<5.4.0",
 ]
 
 [project.scripts]
@@ -63,7 +62,7 @@ members = ["rust/exo_pyo3_bindings", "bench"]
 [tool.uv.sources]
 exo_pyo3_bindings = { workspace = true }
 mlx = { git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git", branch = "address-rdma-gpu-locks", marker = "sys_platform == 'darwin'" }
-mlx-lm = { git = "https://github.com/rltakashige/mlx-lm", branch = "leo/fix-arrayscache-leak" }
+mlx-lm = { git = "https://github.com/rltakashige/mlx-lm", branch = "fix/float32-logprobs" }
 # Uncomment to use local mlx/mlx-lm development versions:
 # mlx = { path = "/Users/Shared/mlx", editable=true }
 # mlx-lm = { path = "/Users/Shared/mlx-lm", editable=true }
@@ -116,7 +115,6 @@ root = "src"
 required-version = ">=0.8.6"
 prerelease = "allow"
 environments = ["sys_platform == 'darwin'", "sys_platform == 'linux'"]
-extra-build-dependencies = { "miniaudio" = ["setuptools", "cffi", "pycparser"] }
 
 ###
 # ruff configuration

resources/inference_model_cards/mlx-community--DeepSeek-V3.1-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/DeepSeek-V3.1-4bit"
 n_layers = 61
 hidden_size = 7168
-num_key_value_heads = 128
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "deepseek"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "DeepSeek V3.1"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 405874409472

resources/inference_model_cards/mlx-community--DeepSeek-V3.1-8bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/DeepSeek-V3.1-8bit"
 n_layers = 61
 hidden_size = 7168
-num_key_value_heads = 128
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "deepseek"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "DeepSeek V3.1"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 765577920512

resources/inference_model_cards/mlx-community--DeepSeek-V3.2-4bit.toml

@@ -1,15 +0,0 @@
-model_id = "mlx-community/DeepSeek-V3.2-4bit"
-n_layers = 61
-hidden_size = 7168
-num_key_value_heads = 128
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "deepseek"
-quantization = "4bit"
-base_model = "DeepSeek V3.2"
-capabilities = ["text", "thinking", "thinking_toggle"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 378086226621

resources/inference_model_cards/mlx-community--DeepSeek-V3.2-8bit.toml

@@ -1,15 +0,0 @@
-model_id = "mlx-community/DeepSeek-V3.2-8bit"
-n_layers = 61
-hidden_size = 7168
-num_key_value_heads = 128
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "deepseek"
-quantization = "8bit"
-base_model = "DeepSeek V3.2"
-capabilities = ["text", "thinking", "thinking_toggle"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 755957120916

resources/inference_model_cards/mlx-community--GLM-4.5-Air-8bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-4.5-Air-8bit"
 n_layers = 46
 hidden_size = 4096
-num_key_value_heads = 8
 supports_tensor = false
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "GLM 4.5 Air"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 122406567936

resources/inference_model_cards/mlx-community--GLM-4.5-Air-bf16.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-4.5-Air-bf16"
 n_layers = 46
 hidden_size = 4096
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "bf16"
 base_model = "GLM 4.5 Air"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 229780750336

resources/inference_model_cards/mlx-community--GLM-4.7-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-4.7-4bit"
 n_layers = 91
 hidden_size = 5120
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "GLM 4.7"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 198556925568

resources/inference_model_cards/mlx-community--GLM-4.7-6bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-4.7-6bit"
 n_layers = 91
 hidden_size = 5120
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "6bit"
 base_model = "GLM 4.7"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 286737579648

resources/inference_model_cards/mlx-community--GLM-4.7-8bit-gs32.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-4.7-8bit-gs32"
 n_layers = 91
 hidden_size = 5120
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "GLM 4.7"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 396963397248

resources/inference_model_cards/mlx-community--GLM-4.7-Flash-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-4.7-Flash-4bit"
 n_layers = 47
 hidden_size = 2048
-num_key_value_heads = 20
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "GLM 4.7 Flash"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 19327352832

resources/inference_model_cards/mlx-community--GLM-4.7-Flash-5bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-4.7-Flash-5bit"
 n_layers = 47
 hidden_size = 2048
-num_key_value_heads = 20
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "5bit"
 base_model = "GLM 4.7 Flash"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 22548578304

resources/inference_model_cards/mlx-community--GLM-4.7-Flash-6bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-4.7-Flash-6bit"
 n_layers = 47
 hidden_size = 2048
-num_key_value_heads = 20
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "6bit"
 base_model = "GLM 4.7 Flash"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 26843545600

resources/inference_model_cards/mlx-community--GLM-4.7-Flash-8bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-4.7-Flash-8bit"
 n_layers = 47
 hidden_size = 2048
-num_key_value_heads = 20
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "GLM 4.7 Flash"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 34359738368

resources/inference_model_cards/mlx-community--GLM-5-8bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-5-8bit-MXFP8"
 n_layers = 78
 hidden_size = 6144
-num_key_value_heads = 64
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "GLM-5"
 capabilities = ["text", "thinking"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 790517400864

resources/inference_model_cards/mlx-community--GLM-5-MXFP4-Q8.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-5-MXFP4-Q8"
 n_layers = 78
 hidden_size = 6144
-num_key_value_heads = 64
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "MXFP4-Q8"
 base_model = "GLM-5"
 capabilities = ["text", "thinking"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 405478939008

resources/inference_model_cards/mlx-community--GLM-5-bf16.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/GLM-5"
 n_layers = 78
 hidden_size = 6144
-num_key_value_heads = 64
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "glm"
@@ -9,7 +8,5 @@ quantization = "bf16"
 base_model = "GLM-5"
 capabilities = ["text", "thinking"]
 
-context_length = 202752
-
 [storage_size]
 in_bytes = 1487822475264

resources/inference_model_cards/mlx-community--Kimi-K2-Instruct-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Kimi-K2-Instruct-4bit"
 n_layers = 61
 hidden_size = 7168
-num_key_value_heads = 64
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "kimi"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "Kimi K2"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 620622774272

resources/inference_model_cards/mlx-community--Kimi-K2-Thinking.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Kimi-K2-Thinking"
 n_layers = 61
 hidden_size = 7168
-num_key_value_heads = 64
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "kimi"
@@ -9,7 +8,5 @@ quantization = ""
 base_model = "Kimi K2"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 262144
-
 [storage_size]
 in_bytes = 706522120192

resources/inference_model_cards/mlx-community--Kimi-K2.5.toml

@@ -1,21 +1,12 @@
 model_id = "mlx-community/Kimi-K2.5"
 n_layers = 61
 hidden_size = 7168
-num_key_value_heads = 64
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "kimi"
 quantization = ""
 base_model = "Kimi K2.5"
-capabilities = ["text", "thinking", "thinking_toggle", "vision"]
-
-context_length = 262144
+capabilities = ["text", "thinking", "thinking_toggle"]
 
 [storage_size]
 in_bytes = 662498705408
-
-[vision]
-image_token_id = 163605
-model_type = "kimi_vl"
-weights_repo = "davehind/Kimi-K2.5-vision"
-processor_repo = "moonshotai/Kimi-K2.5"

resources/inference_model_cards/mlx-community--Llama-3.1-Nemotron-70B-Instruct-HF-4bit.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/Llama-3.1-Nemotron-70B-Instruct-HF-4bit"
-n_layers = 80
-hidden_size = 8192
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "llama"
-quantization = "4bit"
-base_model = "NVIDIA Llama-3.1-Nemotron-70B-Instruct"
-capabilities = ["text"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 39688355840

resources/inference_model_cards/mlx-community--Llama-3.1-Nemotron-70B-Instruct-HF-8bit.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/Llama-3.1-Nemotron-70B-Instruct-HF-8bit"
-n_layers = 80
-hidden_size = 8192
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "llama"
-quantization = "8bit"
-base_model = "NVIDIA Llama-3.1-Nemotron-70B-Instruct"
-capabilities = ["text"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 74964549632

resources/inference_model_cards/mlx-community--Llama-3.1-Nemotron-70B-Instruct-HF-bf16.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/Llama-3.1-Nemotron-70B-Instruct-HF-bf16"
-n_layers = 80
-hidden_size = 8192
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "llama"
-quantization = "bf16"
-base_model = "NVIDIA Llama-3.1-Nemotron-70B-Instruct"
-capabilities = ["text"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 141107412992

resources/inference_model_cards/mlx-community--Llama-3.1-Nemotron-Nano-4B-v1.1-4bit.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/Llama-3.1-Nemotron-Nano-4B-v1.1-4bit"
-n_layers = 32
-hidden_size = 3072
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "llama"
-quantization = "4bit"
-base_model = "NVIDIA Llama-3.1-Nemotron-Nano-4B-v1.1"
-capabilities = ["text"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 2538706944

resources/inference_model_cards/mlx-community--Llama-3.1-Nemotron-Nano-4B-v1.1-8bit.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/Llama-3.1-Nemotron-Nano-4B-v1.1-8bit"
-n_layers = 32
-hidden_size = 3072
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "llama"
-quantization = "8bit"
-base_model = "NVIDIA Llama-3.1-Nemotron-Nano-4B-v1.1"
-capabilities = ["text"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 4794980352

resources/inference_model_cards/mlx-community--Llama-3.1-Nemotron-Nano-4B-v1.1-bf16.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/Llama-3.1-Nemotron-Nano-4B-v1.1-bf16"
-n_layers = 32
-hidden_size = 3072
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "llama"
-quantization = "bf16"
-base_model = "NVIDIA Llama-3.1-Nemotron-Nano-4B-v1.1"
-capabilities = ["text"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 9025492992

resources/inference_model_cards/mlx-community--Llama-3.2-1B-Instruct-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Llama-3.2-1B-Instruct-4bit"
 n_layers = 16
 hidden_size = 2048
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "llama"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "Llama 3.2 1B"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 729808896

resources/inference_model_cards/mlx-community--Llama-3.2-3B-Instruct-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Llama-3.2-3B-Instruct-4bit"
 n_layers = 28
 hidden_size = 3072
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "llama"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "Llama 3.2 3B"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 1863319552

resources/inference_model_cards/mlx-community--Llama-3.2-3B-Instruct-8bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Llama-3.2-3B-Instruct-8bit"
 n_layers = 28
 hidden_size = 3072
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "llama"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "Llama 3.2 3B"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 3501195264

resources/inference_model_cards/mlx-community--Llama-3.3-70B-Instruct-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Llama-3.3-70B-Instruct-4bit"
 n_layers = 80
 hidden_size = 8192
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "llama"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "Llama 3.3 70B"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 40652242944

resources/inference_model_cards/mlx-community--Llama-3.3-70B-Instruct-8bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Llama-3.3-70B-Instruct-8bit"
 n_layers = 80
 hidden_size = 8192
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "llama"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "Llama 3.3 70B"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 76799803392

resources/inference_model_cards/mlx-community--Meta-Llama-3.1-70B-Instruct-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Meta-Llama-3.1-70B-Instruct-4bit"
 n_layers = 80
 hidden_size = 8192
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "llama"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "Llama 3.1 70B"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 40652242944

resources/inference_model_cards/mlx-community--Meta-Llama-3.1-8B-Instruct-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Meta-Llama-3.1-8B-Instruct-4bit"
 n_layers = 32
 hidden_size = 4096
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "llama"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "Llama 3.1 8B"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 4637851648

resources/inference_model_cards/mlx-community--Meta-Llama-3.1-8B-Instruct-8bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Meta-Llama-3.1-8B-Instruct-8bit"
 n_layers = 32
 hidden_size = 4096
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "llama"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "Llama 3.1 8B"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 8954839040

resources/inference_model_cards/mlx-community--Meta-Llama-3.1-8B-Instruct-bf16.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Meta-Llama-3.1-8B-Instruct-bf16"
 n_layers = 32
 hidden_size = 4096
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "llama"
@@ -9,7 +8,5 @@ quantization = "bf16"
 base_model = "Llama 3.1 8B"
 capabilities = ["text"]
 
-context_length = 131072
-
 [storage_size]
 in_bytes = 16882073600

resources/inference_model_cards/mlx-community--MiniMax-M2.1-3bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/MiniMax-M2.1-3bit"
 n_layers = 61
 hidden_size = 3072
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "minimax"
@@ -9,7 +8,5 @@ quantization = "3bit"
 base_model = "MiniMax M2.1"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 196608
-
 [storage_size]
 in_bytes = 100086644736

resources/inference_model_cards/mlx-community--MiniMax-M2.1-8bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/MiniMax-M2.1-8bit"
 n_layers = 61
 hidden_size = 3072
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "minimax"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "MiniMax M2.1"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 196608
-
 [storage_size]
 in_bytes = 242986745856

resources/inference_model_cards/mlx-community--MiniMax-M2.5-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/MiniMax-M2.5-4bit"
 n_layers = 62
 hidden_size = 3072
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "minimax"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "MiniMax M2.5"
 capabilities = ["text", "thinking"]
 
-context_length = 196608
-
 [storage_size]
 in_bytes = 128666664960

resources/inference_model_cards/mlx-community--MiniMax-M2.5-6bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/MiniMax-M2.5-6bit"
 n_layers = 62
 hidden_size = 3072
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "minimax"
@@ -9,7 +8,5 @@ quantization = "6bit"
 base_model = "MiniMax M2.5"
 capabilities = ["text", "thinking"]
 
-context_length = 196608
-
 [storage_size]
 in_bytes = 185826705408

resources/inference_model_cards/mlx-community--MiniMax-M2.5-8bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/MiniMax-M2.5-8bit"
 n_layers = 62
 hidden_size = 3072
-num_key_value_heads = 8
 supports_tensor = true
 tasks = ["TextGeneration"]
 family = "minimax"
@@ -9,7 +8,5 @@ quantization = "8bit"
 base_model = "MiniMax M2.5"
 capabilities = ["text", "thinking"]
 
-context_length = 196608
-
 [storage_size]
 in_bytes = 242986745856

resources/inference_model_cards/mlx-community--NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-4Bit.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-4Bit"
-n_layers = 52
-hidden_size = 2688
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "nemotron"
-quantization = "4bit"
-base_model = "NVIDIA Nemotron-3-Nano-30B-A3B"
-capabilities = ["text"]
-
-context_length = 262144
-
-[storage_size]
-in_bytes = 17775342336

resources/inference_model_cards/mlx-community--NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-5Bit.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-5Bit"
-n_layers = 52
-hidden_size = 2688
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "nemotron"
-quantization = "5bit"
-base_model = "NVIDIA Nemotron-3-Nano-30B-A3B"
-capabilities = ["text"]
-
-context_length = 262144
-
-[storage_size]
-in_bytes = 21721476864

resources/inference_model_cards/mlx-community--NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-6Bit.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-6Bit"
-n_layers = 52
-hidden_size = 2688
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "nemotron"
-quantization = "6bit"
-base_model = "NVIDIA Nemotron-3-Nano-30B-A3B"
-capabilities = ["text"]
-
-context_length = 262144
-
-[storage_size]
-in_bytes = 25667611392

resources/inference_model_cards/mlx-community--NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-8Bit.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-8Bit"
-n_layers = 52
-hidden_size = 2688
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "nemotron"
-quantization = "8bit"
-base_model = "NVIDIA Nemotron-3-Nano-30B-A3B"
-capabilities = ["text"]
-
-context_length = 262144
-
-[storage_size]
-in_bytes = 33559880448

resources/inference_model_cards/mlx-community--NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-BF16.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-BF16"
-n_layers = 52
-hidden_size = 2688
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "nemotron"
-quantization = "bf16"
-base_model = "NVIDIA Nemotron-3-Nano-30B-A3B"
-capabilities = ["text"]
-
-context_length = 262144
-
-[storage_size]
-in_bytes = 63155889408

resources/inference_model_cards/mlx-community--NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-MXFP4.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-MXFP4"
-n_layers = 52
-hidden_size = 2688
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "nemotron"
-quantization = "4bit"
-base_model = "NVIDIA Nemotron-3-Nano-30B-A3B"
-capabilities = ["text"]
-
-context_length = 262144
-
-[storage_size]
-in_bytes = 16788808704

resources/inference_model_cards/mlx-community--NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4"
-n_layers = 52
-hidden_size = 2688
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "nemotron"
-quantization = "4bit"
-base_model = "NVIDIA Nemotron-3-Nano-30B-A3B"
-capabilities = ["text"]
-
-context_length = 262144
-
-[storage_size]
-in_bytes = 19323906944

resources/inference_model_cards/mlx-community--NVIDIA-Nemotron-Nano-9B-v2-4bits.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/NVIDIA-Nemotron-Nano-9B-v2-4bits"
-n_layers = 56
-hidden_size = 4480
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "nemotron"
-quantization = "4bit"
-base_model = "NVIDIA Nemotron-Nano-9B-v2"
-capabilities = ["text"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 5002791936

resources/inference_model_cards/mlx-community--NVIDIA-Nemotron-Nano-9B-v2-6bit.toml

@@ -1,14 +0,0 @@
-model_id = "mlx-community/NVIDIA-Nemotron-Nano-9B-v2-6bit"
-n_layers = 56
-hidden_size = 4480
-supports_tensor = true
-tasks = ["TextGeneration"]
-family = "nemotron"
-quantization = "6bit"
-base_model = "NVIDIA Nemotron-Nano-9B-v2"
-capabilities = ["text"]
-
-context_length = 131072
-
-[storage_size]
-in_bytes = 7224298496

resources/inference_model_cards/mlx-community--Qwen3-0.6B-4bit.toml

@@ -1,7 +1,6 @@
 model_id = "mlx-community/Qwen3-0.6B-4bit"
 n_layers = 28
 hidden_size = 1024
-num_key_value_heads = 8
 supports_tensor = false
 tasks = ["TextGeneration"]
 family = "qwen"
@@ -9,7 +8,5 @@ quantization = "4bit"
 base_model = "Qwen3 0.6B"
 capabilities = ["text", "thinking", "thinking_toggle"]
 
-context_length = 32768
-
 [storage_size]
 in_bytes = 342884352