chore: apply nix fmt formatting to svelte files

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
feat: add light/dark mode toggle to dashboard
2026-02-18 11:46:00 -08:00 · 2026-02-18 11:04:09 -08:00 · 2026-02-18 17:05:00 +00:00 · 2026-02-18 16:18:09 +00:00 · 2026-02-18 16:05:39 +00:00 · 2026-02-18 14:04:06 +00:00
71 changed files with 1392 additions and 708 deletions
@@ -0,0 +1,46 @@
+"""Type stubs for mlx_lm.models.glm_moe_dsa"""
+
+from dataclasses import dataclass
+from typing import Any, Dict, Optional
+
+from .base import BaseModelArgs
+from .deepseek_v32 import Model as DSV32Model
+
+@dataclass
+class ModelArgs(BaseModelArgs):
+    model_type: str
+    vocab_size: int
+    hidden_size: int
+    index_head_dim: int
+    index_n_heads: int
+    index_topk: int
+    intermediate_size: int
+    moe_intermediate_size: int
+    num_hidden_layers: int
+    num_attention_heads: int
+    num_key_value_heads: int
+    n_shared_experts: Optional[int]
+    n_routed_experts: Optional[int]
+    routed_scaling_factor: float
+    kv_lora_rank: int
+    q_lora_rank: int
+    qk_rope_head_dim: int
+    v_head_dim: int
+    qk_nope_head_dim: int
+    topk_method: str
+    scoring_func: str
+    norm_topk_prob: bool
+    n_group: int
+    topk_group: int
+    num_experts_per_tok: int
+    moe_layer_freq: int
+    first_k_dense_replace: int
+    max_position_embeddings: int
+    rms_norm_eps: float
+    rope_parameters: Dict[str, Any]
+    attention_bias: bool
+    rope_scaling: Dict[str, Any] | None
+    rope_theta: float | None
+
+class Model(DSV32Model):
+    def __init__(self, config: ModelArgs) -> None: ...
@@ -141,12 +141,6 @@ version = "0.3.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "76a2e8124351fda1ef8aaaa3bbd7ebbcb486bbcd4225aca0aa0d84bb2db8fecb"

-[[package]]
-name = "arrayvec"
-version = "0.7.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7c02d123df017efcdfbd739ef81735b36c5ba83ec3c59c80a9d7ecc718f92e50"
-
 [[package]]
 name = "asn1-rs"
 version = "0.7.1"
@@ -304,19 +298,6 @@ version = "1.8.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "55248b47b0caf0546f7988906588779981c43bb1bc9d0c44087278f80cdb44ba"

-[[package]]
-name = "bigdecimal"
-version = "0.4.9"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "560f42649de9fa436b73517378a147ec21f6c997a546581df4b4b31677828934"
-dependencies = [
- "autocfg",
- "libm",
- "num-bigint",
- "num-integer",
- "num-traits",
-]
-
 [[package]]
 name = "bimap"
 version = "0.6.3"
@@ -516,15 +497,6 @@ version = "0.4.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "2f421161cb492475f1661ddc9815a745a1c894592070661180fdec3d4872e9c3"

-[[package]]
-name = "convert_case"
-version = "0.10.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "633458d4ef8c78b72454de2d54fd6ab2e60f9e02be22f3c6104cdc8a4e0fceb9"
-dependencies = [
- "unicode-segmentation",
-]
-
 [[package]]
 name = "core-foundation"
 version = "0.9.4"
@@ -746,29 +718,6 @@ dependencies = [
 "powerfmt",
 ]

-[[package]]
-name = "derive_more"
-version = "2.1.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "10b768e943bed7bf2cab53df09f4bc34bfd217cdb57d971e769874c9a6710618"
-dependencies = [
- "derive_more-impl",
-]
-
-[[package]]
-name = "derive_more-impl"
-version = "2.1.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6d286bfdaf75e988b4a78e013ecd79c581e06399ab53fbacd2d916c2f904f30b"
-dependencies = [
- "convert_case",
- "proc-macro2",
- "quote",
- "rustc_version",
- "syn 2.0.111",
- "unicode-xid",
-]
-
 [[package]]
 name = "digest"
 version = "0.10.7"
@@ -939,22 +888,17 @@ name = "exo_pyo3_bindings"
 version = "0.0.1"
 dependencies = [
 "delegate",
- "derive_more",
 "env_logger",
 "extend",
 "futures",
- "impl-trait-for-tuples",
 "libp2p",
 "log",
 "networking",
- "once_cell",
 "pin-project",
 "pyo3",
 "pyo3-async-runtimes",
 "pyo3-log",
 "pyo3-stub-gen",
- "thiserror 2.0.17",
- "thread_local",
 "tokio",
 "util",
 ]
@@ -1640,17 +1584,6 @@ dependencies = [
 "xmltree",
 ]

-[[package]]
-name = "impl-trait-for-tuples"
-version = "0.2.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a0eb5a3343abf848c0984fe4604b2b105da9539376e24fc0a3b0007411ae4fd9"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn 2.0.111",
-]
-
 [[package]]
 name = "indexmap"
 version = "2.12.1"
@@ -1829,12 +1762,6 @@ version = "0.2.178"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "37c93d8daa9d8a012fd8ab92f088405fb202ea0b6ab73ee2482ae66af4f42091"

-[[package]]
-name = "libm"
-version = "0.2.15"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f9fbbcab51052fe104eb5e5d351cf728d30a5be1fe14d9be8a3b097481fb97de"
-
 [[package]]
 name = "libp2p"
 version = "0.56.0"
@@ -2824,16 +2751,13 @@ name = "networking"
 version = "0.0.1"
 dependencies = [
 "delegate",
- "derive_more",
 "either",
 "extend",
 "futures",
 "futures-timer",
- "impl-trait-for-tuples",
 "keccak-const",
 "libp2p",
 "log",
- "thiserror 2.0.17",
 "tokio",
 "tracing-subscriber",
 "util",
@@ -2918,17 +2842,6 @@ dependencies = [
 "num-traits",
 ]

-[[package]]
-name = "num-rational"
-version = "0.4.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f83d14da390562dca69fc84082e73e548e1ad308d24accdedd2720017cb37824"
-dependencies = [
- "num-bigint",
- "num-integer",
- "num-traits",
-]
-
 [[package]]
 name = "num-traits"
 version = "0.2.19"
@@ -3279,28 +3192,14 @@ version = "0.27.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ab53c047fcd1a1d2a8820fe84f05d6be69e9526be40cb03b73f86b6b03e6d87d"
 dependencies = [
- "bigdecimal",
- "either",
- "hashbrown 0.16.1",
- "indexmap",
 "indoc",
- "inventory",
 "libc",
- "lock_api",
 "memoffset",
- "num-bigint",
- "num-complex",
- "num-rational",
- "num-traits",
 "once_cell",
- "ordered-float",
- "parking_lot",
 "portable-atomic",
 "pyo3-build-config",
 "pyo3-ffi",
 "pyo3-macros",
- "rust_decimal",
- "smallvec",
 "unindent",
 ]

@@ -3741,16 +3640,6 @@ dependencies = [
 "tokio",
 ]

-[[package]]
-name = "rust_decimal"
-version = "1.39.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "35affe401787a9bd846712274d97654355d21b2a2c092a3139aabe31e9022282"
-dependencies = [
- "arrayvec",
- "num-traits",
-]
-
 [[package]]
 name = "rustc-hash"
 version = "1.1.0"
@@ -4615,24 +4504,12 @@ version = "1.0.22"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9312f7c4f6ff9069b165498234ce8be658059c6728633667c526e27dc2cf1df5"

-[[package]]
-name = "unicode-segmentation"
-version = "1.12.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f6ccf251212114b54433ec949fd6a7841275f9ada20dddd2f29e9ceea4501493"
-
 [[package]]
 name = "unicode-width"
 version = "0.2.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b4ac048d71ede7ee76d585517add45da530660ef4390e49b098733c6e897f254"

-[[package]]
-name = "unicode-xid"
-version = "0.2.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ebc1c04c71510c7f702b52b7c350734c9ff1295c464a03335b00bb84fc54f853"
-
 [[package]]
 name = "unicode_names2"
 version = "1.3.0"
@@ -26,49 +26,21 @@ opt-level = 3
 networking = { path = "rust/networking" }
 util = { path = "rust/util" }

-# Proc-macro authoring tools
-syn = "2.0"
-quote = "1.0"
-proc-macro2 = "1.0"
-darling = "0.20"
-
 # Macro dependecies
 extend = "1.2"
 delegate = "0.13"
-impl-trait-for-tuples = "0.2"
-clap = "4.5"
-derive_more = { version = "2.0.1", features = ["display"] }
 pin-project = "1"

 # Utility dependencies
-itertools = "0.14"
-thiserror = "2"
-internment = "0.8"
-recursion = "0.5"
-regex = "1.11"
-once_cell = "1.21"
-thread_local = "1.1"
-bon = "3.4"
-generativity = "1.1"
-anyhow = "1.0"
 keccak-const = "0.2"

-# Functional generics/lenses frameworks
-frunk_core = "0.4"
-frunk = "0.4"
-frunk_utils = "0.2"
-frunk-enum-core = "0.3"
-
 # Async dependencies
 tokio = "1.46"
 futures = "0.3"
-futures-util = "0.3"
 futures-timer = "3.0"

 # Data structures
 either = "1.15"
-ordered-float = "5.0"
-ahash = "0.8"

 # Tracing/logging
 log = "0.4"
@@ -16,8 +16,8 @@
 	/* Gotham-inspired accent colors */
 	--exo-grid: oklch(0.25 0 0);
 	--exo-scanline: oklch(0.15 0 0);
-	--exo-glow-yellow: 0 0 20px oklch(0.85 0.18 85 / 0.3);
-	--exo-glow-yellow-strong: 0 0 40px oklch(0.85 0.18 85 / 0.5);
+	--exo-glow-yellow: oklch(0.85 0.18 85 / 0.3);
+	--exo-glow-yellow-strong: oklch(0.85 0.18 85 / 0.5);
 	
 	/* Theme Variables */
 	--radius: 0.375rem;
@@ -39,6 +39,67 @@
 	--border: oklch(0.22 0 0);
 	--input: oklch(0.22 0 0);
 	--ring: var(--exo-yellow);
+
+	/* Adaptive text colors (used via utility classes below) */
+	--text-primary: oklch(0.9 0 0);
+	--text-secondary: oklch(0.9 0 0 / 0.9);
+	--text-muted: oklch(0.9 0 0 / 0.7);
+	--text-faint: oklch(0.9 0 0 / 0.5);
+	--text-ghost: oklch(0.9 0 0 / 0.4);
+	--text-subtle: oklch(0.9 0 0 / 0.6);
+	--text-dim: oklch(0.9 0 0 / 0.3);
+	--text-whisper: oklch(0.9 0 0 / 0.8);
+	--exo-bg-hover: oklch(0.18 0 0);
+}
+
+/* ============================================================
+   LIGHT THEME - Applied when html does NOT have .dark class
+   ============================================================ */
+html:not(.dark) {
+	/* Override EXO brand colors for light mode */
+	--exo-black: oklch(0.97 0 0);
+	--exo-dark-gray: oklch(0.94 0 0);
+	--exo-medium-gray: oklch(0.88 0 0);
+	--exo-light-gray: oklch(0.45 0 0);
+	--exo-yellow: oklch(0.20 0.02 85);
+	--exo-yellow-darker: oklch(0.15 0.02 85);
+	--exo-yellow-glow: oklch(0.30 0.03 85);
+
+	--exo-grid: oklch(0.88 0 0);
+	--exo-scanline: oklch(0.92 0 0);
+	--exo-glow-yellow: oklch(0.20 0.02 85 / 0.08);
+	--exo-glow-yellow-strong: oklch(0.20 0.02 85 / 0.12);
+
+	/* Theme Variables */
+	--background: oklch(0.97 0 0);
+	--foreground: oklch(0.15 0 0);
+	--card: oklch(0.94 0 0);
+	--card-foreground: oklch(0.15 0 0);
+	--popover: oklch(0.96 0 0);
+	--popover-foreground: oklch(0.15 0 0);
+	--primary: oklch(0.20 0.02 85);
+	--primary-foreground: oklch(0.97 0 0);
+	--secondary: oklch(0.90 0 0);
+	--secondary-foreground: oklch(0.15 0 0);
+	--muted: oklch(0.92 0 0);
+	--muted-foreground: oklch(0.45 0 0);
+	--accent: oklch(0.90 0 0);
+	--accent-foreground: oklch(0.15 0 0);
+	--destructive: oklch(0.55 0.25 25);
+	--border: oklch(0.85 0 0);
+	--input: oklch(0.88 0 0);
+	--ring: oklch(0.20 0.02 85);
+
+	/* Adaptive text colors for light mode */
+	--text-primary: oklch(0.15 0 0);
+	--text-secondary: oklch(0.15 0 0 / 0.9);
+	--text-muted: oklch(0.15 0 0 / 0.7);
+	--text-faint: oklch(0.15 0 0 / 0.5);
+	--text-ghost: oklch(0.15 0 0 / 0.4);
+	--text-subtle: oklch(0.15 0 0 / 0.6);
+	--text-dim: oklch(0.15 0 0 / 0.3);
+	--text-whisper: oklch(0.15 0 0 / 0.8);
+	--exo-bg-hover: oklch(0.90 0 0);
 }

@theme inline {
@@ -266,6 +327,327 @@ input:focus, textarea:focus {
 	box-shadow: none;
 }

+/* ============================================================
+   LIGHT MODE: Override hardcoded white/black text colors
+   Components use text-white/XX extensively - in light mode these
+   need to become dark text. We override at the CSS level so we
+   don't have to touch every single component.
+   ============================================================ */
+html:not(.dark) {
+	/* text-white variants → dark text */
+	& .text-white\/90,
+	& .text-white\/80,
+	& .text-white\/70,
+	& .text-white\/60 {
+		color: var(--foreground) !important;
+	}
+
+	/* Muted text-white variants → dark text with reduced alpha via color-mix */
+	& .text-white\/50 {
+		color: color-mix(in oklch, var(--foreground) 50%, transparent) !important;
+	}
+	& .text-white\/40 {
+		color: color-mix(in oklch, var(--foreground) 40%, transparent) !important;
+	}
+	& .text-white\/30 {
+		color: color-mix(in oklch, var(--foreground) 30%, transparent) !important;
+	}
+
+	/* bg-black variants → light bg */
+	& .bg-black\/40,
+	& .bg-black\/50,
+	& .bg-black\/60,
+	& .bg-black\/80 {
+		background-color: oklch(0.92 0 0 / 0.5) !important;
+	}
+
+	/* bg-exo-black opacity variants */
+	& [class*="bg-exo-black/"] {
+		background-color: oklch(0.92 0 0 / 0.5) !important;
+	}
+
+	/* shadow-black → softer shadow */
+	& [class*="shadow-black"] {
+		--tw-shadow-color: oklch(0 0 0 / 0.08) !important;
+	}
+
+	/* Scrollbar light mode */
+	& ::-webkit-scrollbar-track {
+		background: oklch(0.94 0 0) !important;
+	}
+	& ::-webkit-scrollbar-thumb {
+		background: oklch(0.78 0 0) !important;
+	}
+	& ::-webkit-scrollbar-thumb:hover {
+		background: oklch(0.40 0 0) !important;
+	}
+
+	/* Command panel light mode */
+	& .command-panel {
+		background: linear-gradient(
+			180deg,
+			oklch(0.96 0 0 / 0.95) 0%,
+			oklch(0.94 0 0 / 0.98) 100%
+		) !important;
+		border-color: oklch(0.85 0 0) !important;
+		box-shadow: 
+			inset 0 1px 0 oklch(1 0 0 / 0.5),
+			0 4px 20px oklch(0 0 0 / 0.08) !important;
+	}
+
+	/* Glow text - no glow in light mode, just subtle shadow */
+	& .glow-text {
+		text-shadow: 
+			0 0 10px oklch(0 0 0 / 0.06),
+			0 0 20px oklch(0 0 0 / 0.03) !important;
+	}
+
+	/* Grid bg */
+	& .grid-bg {
+		background-image: 
+			linear-gradient(oklch(0.85 0 0 / 0.3) 1px, transparent 1px),
+			linear-gradient(90deg, oklch(0.85 0 0 / 0.3) 1px, transparent 1px) !important;
+	}
+
+	/* Scanlines - very subtle in light mode */
+	& .scanlines::before {
+		background: repeating-linear-gradient(
+			0deg,
+			transparent,
+			transparent 2px,
+			oklch(0 0 0 / 0.015) 2px,
+			oklch(0 0 0 / 0.015) 4px
+		) !important;
+	}
+
+	/* CRT screen effect */
+	& .crt-screen {
+		background: radial-gradient(
+			ellipse at center,
+			oklch(0.96 0 0) 0%,
+			oklch(0.94 0 0) 50%,
+			oklch(0.92 0 0) 100%
+		) !important;
+		box-shadow:
+			inset 0 0 100px oklch(0 0 0 / 0.05),
+			0 0 50px oklch(0 0 0 / 0.03) !important;
+	}
+
+	/* Graph links */
+	& .graph-link {
+		stroke: oklch(0.35 0 0 / 0.5) !important;
+		filter: none !important;
+	}
+
+	& .graph-link-active {
+		stroke: oklch(0.20 0 0 / 0.8) !important;
+		filter: none !important;
+	}
+
+	/* Shooting stars - hide in light mode */
+	& .shooting-stars {
+		display: none !important;
+	}
+
+	/* Logo: invert white→black + subtle dark shadow */
+	& img[alt="EXO"] {
+		filter: invert(1) drop-shadow(0 0 10px oklch(0 0 0 / 0.15)) !important;
+	}
+
+	/* Status indicator colors */
+	& .bg-exo-yellow.rounded-full {
+		box-shadow: 0 0 6px oklch(0 0 0 / 0.15);
+	}
+
+	/* Red/green/blue semantic colors stay mostly the same but darken slightly */
+	& .text-red-400 {
+		color: oklch(0.55 0.22 25) !important;
+	}
+	& .text-green-400 {
+		color: oklch(0.50 0.18 155) !important;
+	}
+	& .text-blue-200 {
+		color: oklch(0.50 0.16 250) !important;
+	}
+	& .text-blue-300 {
+		color: oklch(0.50 0.17 250) !important;
+	}
+	& .text-blue-400 {
+		color: oklch(0.50 0.18 250) !important;
+	}
+
+	/* bg-red variants */
+	& .bg-red-500\/10 {
+		background-color: oklch(0.55 0.22 25 / 0.08) !important;
+	}
+	& .bg-red-500\/20 {
+		background-color: oklch(0.55 0.22 25 / 0.12) !important;
+	}
+	& .bg-red-500\/30 {
+		background-color: oklch(0.55 0.22 25 / 0.15) !important;
+	}
+
+	/* Input/textarea backgrounds */
+	& textarea,
+	& input[type="text"] {
+		color: var(--foreground) !important;
+	}
+	& textarea::placeholder,
+	& input::placeholder {
+		color: oklch(0.45 0 0 / 0.5) !important;
+	}
+
+	/* KaTeX / Math colors in light mode */
+	& .katex,
+	& .katex .mord,
+	& .katex .minner,
+	& .katex .mop,
+	& .katex .mbin,
+	& .katex .mrel,
+	& .katex .mpunct {
+		color: oklch(0.15 0 0) !important;
+	}
+
+	& .katex .frac-line,
+	& .katex .overline-line,
+	& .katex .underline-line,
+	& .katex .hline,
+	& .katex .rule {
+		border-color: oklch(0.25 0 0) !important;
+		background: oklch(0.25 0 0) !important;
+	}
+
+	& .katex .sqrt-line {
+		border-color: oklch(0.25 0 0) !important;
+	}
+
+	& .katex svg {
+		fill: oklch(0.25 0 0) !important;
+		stroke: oklch(0.25 0 0) !important;
+	}
+
+	& .katex svg path {
+		stroke: oklch(0.25 0 0) !important;
+	}
+
+	& .katex .delimsizing,
+	& .katex .delim-size1,
+	& .katex .delim-size2,
+	& .katex .delim-size3,
+	& .katex .delim-size4,
+	& .katex .mopen,
+	& .katex .mclose {
+		color: oklch(0.35 0 0) !important;
+	}
+
+	/* Code blocks in light mode */
+	& .hljs {
+		color: oklch(0.25 0 0) !important;
+	}
+
+	& .code-block-wrapper,
+	& .math-display-wrapper {
+		background: oklch(0.96 0 0) !important;
+		border-color: oklch(0.85 0 0) !important;
+	}
+
+	& .code-block-header,
+	& .math-display-header {
+		background: oklch(0.93 0 0) !important;
+		border-color: oklch(0.87 0 0) !important;
+	}
+
+	/* Inline code */
+	& .inline-code {
+		background: oklch(0.90 0 0) !important;
+		color: oklch(0.25 0 0) !important;
+	}
+
+	/* Blockquote */
+	& blockquote {
+		background: oklch(0.94 0 0) !important;
+	}
+
+	/* Table */
+	& th {
+		background: oklch(0.92 0 0) !important;
+		border-color: oklch(0.82 0 0) !important;
+	}
+	& td {
+		border-color: oklch(0.85 0 0) !important;
+	}
+	& hr {
+		border-color: oklch(0.85 0 0) !important;
+	}
+
+	/* hljs syntax colors for light mode */
+	& .hljs-keyword,
+	& .hljs-selector-tag,
+	& .hljs-literal,
+	& .hljs-section,
+	& .hljs-link {
+		color: oklch(0.50 0.25 300) !important;
+	}
+
+	& .hljs-string,
+	& .hljs-title,
+	& .hljs-name,
+	& .hljs-type,
+	& .hljs-attribute,
+	& .hljs-symbol,
+	& .hljs-bullet,
+	& .hljs-addition,
+	& .hljs-variable,
+	& .hljs-template-tag,
+	& .hljs-template-variable {
+		color: oklch(0.45 0.12 55) !important;
+	}
+
+	& .hljs-comment,
+	& .hljs-quote,
+	& .hljs-deletion,
+	& .hljs-meta {
+		color: oklch(0.55 0 0) !important;
+	}
+
+	& .hljs-number,
+	& .hljs-regexp,
+	& .hljs-built_in {
+		color: oklch(0.45 0.16 160) !important;
+	}
+
+	& .hljs-function,
+	& .hljs-class .hljs-title {
+		color: oklch(0.45 0.18 250) !important;
+	}
+
+	/* LaTeX proof/theorem blocks */
+	& .latex-proof {
+		background: oklch(0.97 0 0) !important;
+		border-left-color: oklch(0.70 0 0) !important;
+	}
+	& .latex-proof-header {
+		color: oklch(0.25 0 0) !important;
+	}
+	& .latex-proof-content {
+		color: oklch(0.15 0 0) !important;
+	}
+	& .latex-proof-content::after {
+		color: oklch(0.45 0 0) !important;
+	}
+
+	& .latex-theorem {
+		background: oklch(0.95 0 0) !important;
+		border-color: oklch(0.82 0 0) !important;
+	}
+
+	& .latex-diagram-placeholder {
+		background: oklch(0.97 0 0) !important;
+		border-color: oklch(0.82 0 0) !important;
+		color: oklch(0.35 0 0) !important;
+	}
+}
+
 /* Shooting Stars Animation */
 .shooting-stars {
 	position: fixed;
@@ -1,5 +1,5 @@
 <!doctype html>
-<html lang="en">
+<html lang="en" class="dark">
 	<head>
 		<meta charset="utf-8" />
 		<link rel="icon" href="%sveltekit.assets%/favicon.ico" />
@@ -39,7 +39,7 @@
  <div class="flex flex-wrap gap-2 mb-3 px-1">
    {#each files as file (file.id)}
      <div
-        class="group relative flex items-center gap-2 bg-exo-dark-gray/80 border border-exo-yellow/30 rounded px-2.5 py-1.5 text-xs font-mono transition-all hover:border-exo-yellow/50 hover:shadow-[0_0_10px_rgba(255,215,0,0.1)]"
+        class="group relative flex items-center gap-2 bg-exo-dark-gray/80 border border-exo-yellow/30 rounded px-2.5 py-1.5 text-xs font-mono transition-all hover:border-exo-yellow/50 hover:shadow-[0_0_10px_var(--exo-glow-yellow)]"
      >
        <!-- File preview or icon -->
        {#if file.preview && getFileCategory(file.type, file.name) === "image"}
@@ -659,7 +659,7 @@
        class="px-2.5 sm:px-4 py-1.5 sm:py-2 rounded text-xs sm:text-xs tracking-[0.1em] sm:tracking-[0.15em] uppercase font-medium transition-all duration-200 whitespace-nowrap
 					{!canSend || loading || isEditOnlyWithoutImage
          ? 'bg-exo-medium-gray/50 text-exo-light-gray cursor-not-allowed'
-          : 'bg-exo-yellow text-exo-black hover:bg-exo-yellow-darker hover:shadow-[0_0_20px_rgba(255,215,0,0.3)]'}"
+          : 'bg-exo-yellow text-exo-black hover:bg-exo-yellow-darker hover:shadow-[0_0_20px_var(--exo-glow-yellow)]'}"
        aria-label={shouldShowEditMode
          ? "Edit image"
          : isImageModel()
@@ -271,7 +271,7 @@
          <!-- Assistant message header -->
          <div class="flex items-center gap-1.5 sm:gap-2 mb-1.5 sm:mb-2">
            <div
-              class="w-1.5 h-1.5 sm:w-2 sm:h-2 bg-exo-yellow rounded-full shadow-[0_0_10px_rgba(255,215,0,0.5)]"
+              class="w-1.5 h-1.5 sm:w-2 sm:h-2 bg-exo-yellow rounded-full shadow-[0_0_10px_var(--exo-glow-yellow)]"
            ></div>
            <span
              class="text-sm sm:text-xs text-exo-yellow tracking-[0.15em] sm:tracking-[0.2em] uppercase font-medium"
@@ -1,11 +1,22 @@
 <script lang="ts">
  import { browser } from "$app/environment";
+  import { toggleMode, mode } from "mode-watcher";

-  export let showHome = true;
-  export let onHome: (() => void) | null = null;
-  export let showSidebarToggle = false;
-  export let sidebarVisible = true;
-  export let onToggleSidebar: (() => void) | null = null;
+  let {
+    showHome = true,
+    onHome = null,
+    showSidebarToggle = false,
+    sidebarVisible = true,
+    onToggleSidebar = null,
+  }: {
+    showHome?: boolean;
+    onHome?: (() => void) | null;
+    showSidebarToggle?: boolean;
+    sidebarVisible?: boolean;
+    onToggleSidebar?: (() => void) | null;
+  } = $props();
+
+  const currentMode = $derived(mode.current);

  function handleHome(): void {
    if (onHome) {
@@ -75,14 +86,58 @@
    <img
      src="/exo-logo.png"
      alt="EXO"
-      class="h-18 drop-shadow-[0_0_20px_rgba(255,215,0,0.5)]"
+      class="h-18 drop-shadow-[0_0_20px_color-mix(in_oklch,var(--exo-yellow)_50%,transparent)]"
    />
  </button>

-  <!-- Right: Home + Downloads -->
+  <!-- Right: Theme Toggle + Home + Downloads -->
  <div
    class="absolute right-6 top-1/2 -translate-y-1/2 flex items-center gap-4"
  >
+    <!-- Theme Toggle Button -->
+    <button
+      onclick={() => toggleMode()}
+      class="p-2 rounded border border-exo-medium-gray/40 hover:border-exo-yellow/50 transition-colors cursor-pointer"
+      title={currentMode === "dark"
+        ? "Switch to light mode"
+        : "Switch to dark mode"}
+      aria-label={currentMode === "dark"
+        ? "Switch to light mode"
+        : "Switch to dark mode"}
+    >
+      {#if currentMode === "dark"}
+        <!-- Sun icon - click to go light -->
+        <svg
+          class="w-4 h-4 text-exo-light-gray hover:text-exo-yellow transition-colors"
+          fill="none"
+          viewBox="0 0 24 24"
+          stroke="currentColor"
+          stroke-width="2"
+        >
+          <circle cx="12" cy="12" r="5" />
+          <path
+            stroke-linecap="round"
+            d="M12 1v2m0 18v2m-9-11H1m22 0h-2m-2.636-6.364l-1.414 1.414M6.05 6.05L4.636 4.636m0 14.728l1.414-1.414m11.314 1.414l1.414 1.414"
+          />
+        </svg>
+      {:else}
+        <!-- Moon icon - click to go dark -->
+        <svg
+          class="w-4 h-4 text-exo-light-gray hover:text-exo-yellow transition-colors"
+          fill="none"
+          viewBox="0 0 24 24"
+          stroke="currentColor"
+          stroke-width="2"
+        >
+          <path
+            stroke-linecap="round"
+            stroke-linejoin="round"
+            d="M21 12.79A9 9 0 1111.21 3a7 7 0 009.79 9.79z"
+          />
+        </svg>
+      {/if}
+    </button>
+
    {#if showHome}
      <button
        onclick={handleHome}
@@ -669,7 +669,7 @@
    width: 12px;
    height: 12px;
    border-radius: 50%;
-    background: #ffd700;
+    background: var(--exo-yellow);
    cursor: pointer;
    border: none;
  }
@@ -678,7 +678,7 @@
    width: 12px;
    height: 12px;
    border-radius: 50%;
-    background: #ffd700;
+    background: var(--exo-yellow);
    cursor: pointer;
    border: none;
  }
@@ -1051,4 +1051,231 @@
    text-transform: uppercase;
    letter-spacing: 0.05em;
  }
+
+  /* ============================================
+     LIGHT MODE OVERRIDES
+     All dark-mode-specific hardcoded colors are
+     overridden below for light backgrounds.
+     ============================================ */
+
+  /* --- Inline code --- */
+  :global(html:not(.dark)) .markdown-content :global(.inline-code) {
+    background: rgba(0, 0, 0, 0.06);
+    color: var(--exo-yellow, #b8860b);
+  }
+
+  /* --- Blockquote --- */
+  :global(html:not(.dark)) .markdown-content :global(blockquote) {
+    background: rgba(0, 0, 0, 0.03);
+  }
+
+  /* --- Tables --- */
+  :global(html:not(.dark)) .markdown-content :global(th) {
+    background: rgba(0, 0, 0, 0.05);
+    border: 1px solid rgba(0, 0, 0, 0.12);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(td) {
+    border: 1px solid rgba(0, 0, 0, 0.1);
+  }
+
+  /* --- Horizontal rule --- */
+  :global(html:not(.dark)) .markdown-content :global(hr) {
+    border-top: 1px solid rgba(0, 0, 0, 0.12);
+  }
+
+  /* --- Code block wrapper --- */
+  :global(html:not(.dark)) .markdown-content :global(.code-block-wrapper) {
+    border: 1px solid rgba(0, 0, 0, 0.12);
+    background: rgba(0, 0, 0, 0.03);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.code-block-header) {
+    background: rgba(0, 0, 0, 0.04);
+    border-bottom: 1px solid rgba(0, 0, 0, 0.08);
+  }
+
+  /* --- Syntax highlighting - light theme --- */
+  :global(html:not(.dark)) .markdown-content :global(.hljs) {
+    color: #1f2937;
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.hljs-keyword),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-selector-tag),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-literal),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-section),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-link) {
+    color: #7c3aed;
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.hljs-string),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-title),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-name),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-type),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-attribute),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-symbol),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-bullet),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-addition),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-variable),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-template-tag),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-template-variable) {
+    color: #b45309;
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.hljs-comment),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-quote),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-deletion),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-meta) {
+    color: #9ca3af;
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.hljs-number),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-regexp),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-literal),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-built_in) {
+    color: #059669;
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.hljs-function),
+  :global(html:not(.dark)) .markdown-content :global(.hljs-class .hljs-title) {
+    color: #2563eb;
+  }
+
+  /* --- KaTeX math - light mode --- */
+  :global(html:not(.dark)) .markdown-content :global(.katex) {
+    color: oklch(0.2 0 0);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.katex .mord),
+  :global(html:not(.dark)) .markdown-content :global(.katex .minner),
+  :global(html:not(.dark)) .markdown-content :global(.katex .mop),
+  :global(html:not(.dark)) .markdown-content :global(.katex .mbin),
+  :global(html:not(.dark)) .markdown-content :global(.katex .mrel),
+  :global(html:not(.dark)) .markdown-content :global(.katex .mpunct) {
+    color: oklch(0.2 0 0);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.katex .frac-line),
+  :global(html:not(.dark)) .markdown-content :global(.katex .overline-line),
+  :global(html:not(.dark)) .markdown-content :global(.katex .underline-line),
+  :global(html:not(.dark)) .markdown-content :global(.katex .hline),
+  :global(html:not(.dark)) .markdown-content :global(.katex .rule) {
+    border-color: oklch(0.25 0 0) !important;
+    background: oklch(0.25 0 0);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.katex .sqrt-line) {
+    border-color: oklch(0.25 0 0) !important;
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.katex svg) {
+    fill: oklch(0.25 0 0);
+    stroke: oklch(0.25 0 0);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.katex svg path) {
+    stroke: oklch(0.25 0 0);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.katex .delimsizing),
+  :global(html:not(.dark)) .markdown-content :global(.katex .delim-size1),
+  :global(html:not(.dark)) .markdown-content :global(.katex .delim-size2),
+  :global(html:not(.dark)) .markdown-content :global(.katex .delim-size3),
+  :global(html:not(.dark)) .markdown-content :global(.katex .delim-size4),
+  :global(html:not(.dark)) .markdown-content :global(.katex .mopen),
+  :global(html:not(.dark)) .markdown-content :global(.katex .mclose) {
+    color: oklch(0.35 0 0);
+  }
+
+  /* --- Display math wrapper --- */
+  :global(html:not(.dark)) .markdown-content :global(.math-display-wrapper) {
+    border: 1px solid rgba(0, 0, 0, 0.1);
+    background: rgba(0, 0, 0, 0.02);
+  }
+
+  :global(html:not(.dark))
+    .markdown-content
+    :global(.math-display-wrapper:hover) {
+    border-color: rgba(0, 0, 0, 0.18);
+    box-shadow: 0 0 12px rgba(0, 0, 0, 0.05);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.math-display-header) {
+    background: rgba(0, 0, 0, 0.03);
+    border-bottom: 1px solid rgba(0, 0, 0, 0.06);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.math-label) {
+    color: rgba(0, 0, 0, 0.5);
+  }
+
+  /* --- Math scrollbar (light mode) --- */
+  :global(html:not(.dark))
+    .markdown-content
+    :global(.math-display-content::-webkit-scrollbar-track) {
+    background: rgba(0, 0, 0, 0.04);
+  }
+
+  :global(html:not(.dark))
+    .markdown-content
+    :global(.math-display-content::-webkit-scrollbar-thumb) {
+    background: rgba(0, 0, 0, 0.15);
+  }
+
+  :global(html:not(.dark))
+    .markdown-content
+    :global(.math-display-content::-webkit-scrollbar-thumb:hover) {
+    background: rgba(0, 0, 0, 0.25);
+  }
+
+  /* --- Inline math hover --- */
+  :global(html:not(.dark)) .markdown-content :global(.math-inline:hover) {
+    background: rgba(0, 0, 0, 0.04);
+  }
+
+  /* --- LaTeX proof environment --- */
+  :global(html:not(.dark)) .markdown-content :global(.latex-proof) {
+    background: rgba(0, 0, 0, 0.02);
+    border-left: 3px solid rgba(0, 0, 0, 0.25);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.latex-proof-header) {
+    color: oklch(0.25 0 0);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.latex-proof-content) {
+    color: oklch(0.2 0 0);
+  }
+
+  :global(html:not(.dark))
+    .markdown-content
+    :global(.latex-proof-content::after) {
+    color: oklch(0.4 0 0);
+  }
+
+  /* --- LaTeX theorem-like environments --- */
+  :global(html:not(.dark)) .markdown-content :global(.latex-theorem) {
+    background: rgba(0, 0, 0, 0.02);
+    border: 1px solid rgba(0, 0, 0, 0.1);
+  }
+
+  :global(html:not(.dark)) .markdown-content :global(.latex-theorem-content) {
+    color: oklch(0.2 0 0);
+  }
+
+  /* --- LaTeX diagram/figure placeholder --- */
+  :global(html:not(.dark))
+    .markdown-content
+    :global(.latex-diagram-placeholder) {
+    background: rgba(0, 0, 0, 0.02);
+    border: 1px dashed rgba(0, 0, 0, 0.2);
+    color: rgba(0, 0, 0, 0.45);
+  }
+
+  /* --- Math error (light mode) --- */
+  :global(html:not(.dark)) .markdown-content :global(.math-error) {
+    color: #dc2626;
+    background: rgba(220, 38, 38, 0.08);
+    border: 1px solid rgba(220, 38, 38, 0.15);
+  }
 </style>
@@ -6,6 +6,42 @@
    TopologyEdge,
  } from "$lib/stores/app.svelte";
  import { debugMode, topologyData } from "$lib/stores/app.svelte";
+  import { browser } from "$app/environment";
+
+  // Theme detection
+  let isDark = $state(true);
+
+  $effect(() => {
+    if (!browser) return;
+    const update = () => {
+      isDark = document.documentElement.classList.contains("dark");
+    };
+    update();
+    const observer = new MutationObserver(update);
+    observer.observe(document.documentElement, {
+      attributes: true,
+      attributeFilter: ["class"],
+    });
+    return () => observer.disconnect();
+  });
+
+  // Theme-aware colors
+  let tc = $derived({
+    accent: isDark ? "#FFD700" : "#1a1a1a",
+    accentDim: isDark ? "rgba(255,215,0,0.1)" : "rgba(26,26,26,0.06)",
+    accentGlow: isDark ? "rgba(255,215,0,0.1)" : "rgba(0,0,0,0.03)",
+    outlineActive: isDark ? "#FFD700" : "#374151",
+    outlineInactive: isDark ? "#4B5563" : "#d1d5db",
+    strokeActive: isDark ? "#FFD700" : "#374151",
+    strokeInactive: isDark ? "#4B5563" : "#d1d5db",
+    caseFill: isDark ? "#0a0a0a" : "#f3f4f6",
+    caseDetail: isDark ? "#374151" : "#d1d5db",
+    errorDot: isDark ? "#f87171" : "#dc2626",
+    statusInactive: isDark ? "#4B5563" : "#9ca3af",
+    connGood: isDark ? "rgba(255,255,255,0.85)" : "rgba(30,30,30,0.85)",
+    connBad: isDark ? "rgba(248,113,113,0.85)" : "rgba(220,38,38,0.85)",
+    scanlineColor: isDark ? "rgba(255,215,0,0.02)" : "rgba(0,0,0,0.01)",
+  });

  interface Props {
    model: { id: string; name?: string; storage_size_megabytes?: number };
@@ -491,7 +527,7 @@
  <div
    class="bg-exo-dark-gray/60 border {canFit
      ? 'border-exo-yellow/20 group-hover:border-exo-yellow/40'
-      : 'border-red-500/20'} p-3 transition-all duration-200 group-hover:shadow-[0_0_15px_rgba(255,215,0,0.1)]"
+      : 'border-red-500/20'} p-3 transition-all duration-200 group-hover:shadow-[0_0_15px_var(--exo-glow-yellow)]"
  >
    <!-- Model Name & Memory Required -->
    <div class="flex items-start justify-between gap-2 mb-2">
@@ -589,7 +625,8 @@
      >
        <!-- Scanline effect -->
        <div
-          class="absolute inset-0 bg-[repeating-linear-gradient(0deg,transparent,transparent_2px,rgba(255,215,0,0.02)_2px,rgba(255,215,0,0.02)_4px)] pointer-events-none"
+          style="background: repeating-linear-gradient(0deg, transparent, transparent 2px, {tc.scanlineColor} 2px, {tc.scanlineColor} 4px)"
+          class="absolute inset-0 pointer-events-none"
        ></div>

        <svg
@@ -670,7 +707,9 @@
                  y1={node.y}
                  x2={node2.x}
                  y2={node2.y}
-                  stroke={node.isUsed && node2.isUsed ? "#FFD700" : "#374151"}
+                  stroke={node.isUsed && node2.isUsed
+                    ? tc.strokeActive
+                    : tc.strokeInactive}
                  stroke-width="1"
                  stroke-dasharray={node.isUsed && node2.isUsed ? "4,2" : "2,4"}
                  opacity={node.isUsed && node2.isUsed ? 0.4 : 0.15}
@@ -706,9 +745,7 @@
                  dominant-baseline="hanging"
                  font-size="6"
                  font-family="SF Mono, Monaco, monospace"
-                  fill={conn.iface
-                    ? "rgba(255,255,255,0.85)"
-                    : "rgba(248,113,113,0.85)"}
+                  fill={conn.iface ? tc.connGood : tc.connBad}
                >
                  {conn.arrow}
                  {isRdma
@@ -725,9 +762,7 @@
                  dominant-baseline="hanging"
                  font-size="6"
                  font-family="SF Mono, Monaco, monospace"
-                  fill={conn.iface
-                    ? "rgba(255,255,255,0.85)"
-                    : "rgba(248,113,113,0.85)"}
+                  fill={conn.iface ? tc.connGood : tc.connBad}
                >
                  {conn.arrow}
                  {isRdma
@@ -746,9 +781,7 @@
                  dominant-baseline="auto"
                  font-size="6"
                  font-family="SF Mono, Monaco, monospace"
-                  fill={conn.iface
-                    ? "rgba(255,255,255,0.85)"
-                    : "rgba(248,113,113,0.85)"}
+                  fill={conn.iface ? tc.connGood : tc.connBad}
                >
                  {conn.arrow}
                  {isRdma
@@ -767,9 +800,7 @@
                  dominant-baseline="auto"
                  font-size="6"
                  font-family="SF Mono, Monaco, monospace"
-                  fill={conn.iface
-                    ? "rgba(255,255,255,0.85)"
-                    : "rgba(248,113,113,0.85)"}
+                  fill={conn.iface ? tc.connGood : tc.connBad}
                >
                  {conn.arrow}
                  {isRdma
@@ -801,7 +832,7 @@
                    height={node.iconSize * 0.65}
                    rx="2"
                    fill="none"
-                    stroke={node.isUsed ? "#FFD700" : "#4B5563"}
+                    stroke={node.isUsed ? tc.strokeActive : tc.outlineInactive}
                    stroke-width="1.5"
                  />
                  <!-- Screen area (memory fill container) -->
@@ -810,7 +841,7 @@
                    y="2"
                    width={node.iconSize - 8}
                    height={node.screenHeight}
-                    fill="#0a0a0a"
+                    fill={tc.caseFill}
                  />
                  <!-- Current memory fill (gray) -->
                  <rect
@@ -818,7 +849,7 @@
                    y={2 + node.screenHeight - node.currentFillHeight}
                    width={node.iconSize - 8}
                    height={node.currentFillHeight}
-                    fill="#374151"
+                    fill={tc.caseDetail}
                  />
                  <!-- New model memory fill (glowing yellow) -->
                  {#if node.modelUsageGB > 0 && node.isUsed}
@@ -830,7 +861,7 @@
                        node.modelFillHeight}
                      width={node.iconSize - 8}
                      height={node.modelFillHeight}
-                      fill="#FFD700"
+                      fill={tc.accent}
                      filter="url(#memGlow-{filterId})"
                      class="animate-pulse-slow"
                    />
@@ -842,7 +873,7 @@
                      0.68} L {node.iconSize - 2} {node.iconSize *
                      0.78} L 2 {node.iconSize * 0.78} Z"
                    fill="none"
-                    stroke={node.isUsed ? "#FFD700" : "#4B5563"}
+                    stroke={node.isUsed ? tc.strokeActive : tc.outlineInactive}
                    stroke-width="1.5"
                  />
                </g>
@@ -859,7 +890,7 @@
                    height={node.iconSize - 4}
                    rx="4"
                    fill="none"
-                    stroke={node.isUsed ? "#FFD700" : "#4B5563"}
+                    stroke={node.isUsed ? tc.strokeActive : tc.outlineInactive}
                    stroke-width="1.5"
                  />
                  <!-- Memory fill background -->
@@ -868,7 +899,7 @@
                    y="4"
                    width={node.iconSize - 8}
                    height={node.iconSize - 8}
-                    fill="#0a0a0a"
+                    fill={tc.caseFill}
                  />
                  <!-- Current memory fill -->
                  <rect
@@ -877,7 +908,7 @@
                      (node.iconSize - 8) * (1 - node.currentPercent / 100)}
                    width={node.iconSize - 8}
                    height={(node.iconSize - 8) * (node.currentPercent / 100)}
-                    fill="#374151"
+                    fill={tc.caseDetail}
                  />
                  <!-- New model memory fill -->
                  {#if node.modelUsageGB > 0 && node.isUsed}
@@ -887,7 +918,7 @@
                      width={node.iconSize - 8}
                      height={(node.iconSize - 8) *
                        ((node.newPercent - node.currentPercent) / 100)}
-                      fill="#FFD700"
+                      fill={tc.accent}
                      filter="url(#memGlow-{filterId})"
                      class="animate-pulse-slow"
                    />
@@ -906,7 +937,7 @@
                    height={node.iconSize * 0.4}
                    rx="3"
                    fill="none"
-                    stroke={node.isUsed ? "#FFD700" : "#4B5563"}
+                    stroke={node.isUsed ? tc.strokeActive : tc.outlineInactive}
                    stroke-width="1.5"
                  />
                  <!-- Memory fill background -->
@@ -915,7 +946,7 @@
                    y={node.iconSize * 0.32}
                    width={node.iconSize - 8}
                    height={node.iconSize * 0.36}
-                    fill="#0a0a0a"
+                    fill={tc.caseFill}
                  />
                  <!-- Current memory fill -->
                  <rect
@@ -924,7 +955,7 @@
                      node.iconSize * 0.36 * (1 - node.currentPercent / 100)}
                    width={node.iconSize - 8}
                    height={node.iconSize * 0.36 * (node.currentPercent / 100)}
-                    fill="#374151"
+                    fill={tc.caseDetail}
                  />
                  <!-- New model memory fill -->
                  {#if node.modelUsageGB > 0 && node.isUsed}
@@ -936,7 +967,7 @@
                      height={node.iconSize *
                        0.36 *
                        ((node.newPercent - node.currentPercent) / 100)}
-                      fill="#FFD700"
+                      fill={tc.accent}
                      filter="url(#memGlow-{filterId})"
                      class="animate-pulse-slow"
                    />
@@ -955,8 +986,8 @@
                      0.75} {node.iconSize /
                      2},{node.iconSize} 0,{node.iconSize *
                      0.75} 0,{node.iconSize * 0.25}"
-                    fill={node.isUsed ? "rgba(255,215,0,0.1)" : "#0a0a0a"}
-                    stroke={node.isUsed ? "#FFD700" : "#4B5563"}
+                    fill={node.isUsed ? tc.accentDim : tc.caseFill}
+                    stroke={node.isUsed ? tc.strokeActive : tc.outlineInactive}
                    stroke-width="1.5"
                  />
                </g>
@@ -970,9 +1001,9 @@
                font-family="SF Mono, Monaco, monospace"
                fill={node.isUsed
                  ? node.newPercent > 90
-                    ? "#f87171"
-                    : "#FFD700"
-                  : "#4B5563"}
+                    ? tc.errorDot
+                    : tc.accent
+                  : tc.statusInactive}
              >
                {node.newPercent.toFixed(0)}%
              </text>
@@ -36,6 +36,84 @@
  const rdmaCtlData = $derived(nodeRdmaCtl());
  const identitiesData = $derived(nodeIdentities());

+  // Theme-aware colors - read from CSS custom properties
+  function getThemeColors() {
+    if (typeof document === "undefined") return getDefaultColors();
+    const style = getComputedStyle(document.documentElement);
+    const isDark = document.documentElement.classList.contains("dark");
+    return {
+      accent:
+        style.getPropertyValue("--exo-yellow").trim() ||
+        (isDark ? "oklch(0.85 0.18 85)" : "oklch(0.20 0.02 85)"),
+      accentRgb: isDark ? "255,215,0" : "30,28,20",
+      deviceCase: isDark ? "#1a1a1a" : "#e8e8e8",
+      deviceCaseDark: isDark ? "#2c2c2c" : "#d4d4d4",
+      deviceScreen: isDark ? "#0a0a12" : "#f0f0f5",
+      deviceScreenFill: isDark ? "rgba(0,20,40,0.9)" : "rgba(240,242,248,0.95)",
+      labelWhite: isDark ? "#FFFFFF" : "#1a1a1a",
+      labelMuted: isDark ? "rgba(179,179,179,0.9)" : "rgba(80,80,80,0.9)",
+      labelDim: isDark ? "rgba(179,179,179,0.7)" : "rgba(100,100,100,0.7)",
+      wireDefault: isDark ? "rgba(179,179,179,0.8)" : "rgba(120,120,120,0.6)",
+      wireBright: isDark ? "rgba(255,255,255,0.9)" : "rgba(30,30,30,0.9)",
+      wireFiltered: isDark ? "rgba(140,140,140,0.6)" : "rgba(160,160,160,0.5)",
+      gridStroke: isDark
+        ? "var(--exo-light-gray, #B3B3B3)"
+        : "var(--exo-light-gray, #888888)",
+      errorText: isDark ? "rgba(248,113,113,0.9)" : "rgba(220,38,38,0.9)",
+      normalText: isDark ? "rgba(255,255,255,0.85)" : "rgba(30,30,30,0.85)",
+      gpuChip: isDark ? "rgba(80, 80, 90, 0.7)" : "rgba(180, 180, 190, 0.7)",
+      detailOverlay: isDark ? "rgba(0,0,0,0.35)" : "rgba(0,0,0,0.08)",
+      deviceShadow: isDark ? "rgba(0,0,0,0.2)" : "rgba(0,0,0,0.06)",
+      deviceHighlight: isDark
+        ? "rgba(255,255,255,0.08)"
+        : "rgba(255,255,255,0.5)",
+      tbActive: isDark ? "rgba(234,179,8,0.9)" : "rgba(30,28,20,0.9)",
+      tbInactive: isDark ? "rgba(100,100,100,0.7)" : "rgba(160,160,160,0.7)",
+      deviceDetail: isDark ? "#374151" : "#c0c0c0",
+    };
+  }
+
+  function getDefaultColors() {
+    return {
+      accent: "oklch(0.85 0.18 85)",
+      accentRgb: "255,215,0",
+      deviceCase: "#1a1a1a",
+      deviceCaseDark: "#2c2c2c",
+      deviceScreen: "#0a0a12",
+      deviceScreenFill: "rgba(0,20,40,0.9)",
+      labelWhite: "#FFFFFF",
+      labelMuted: "rgba(179,179,179,0.9)",
+      labelDim: "rgba(179,179,179,0.7)",
+      wireDefault: "rgba(179,179,179,0.8)",
+      wireBright: "rgba(255,255,255,0.9)",
+      wireFiltered: "rgba(140,140,140,0.6)",
+      gridStroke: "var(--exo-light-gray, #B3B3B3)",
+      errorText: "rgba(248,113,113,0.9)",
+      normalText: "rgba(255,255,255,0.85)",
+      gpuChip: "rgba(80, 80, 90, 0.7)",
+      detailOverlay: "rgba(0,0,0,0.35)",
+      deviceShadow: "rgba(0,0,0,0.2)",
+      deviceHighlight: "rgba(255,255,255,0.08)",
+      tbActive: "rgba(234,179,8,0.9)",
+      tbInactive: "rgba(100,100,100,0.7)",
+      deviceDetail: "#374151",
+    };
+  }
+
+  let themeColors = $state(getThemeColors());
+
+  $effect(() => {
+    if (typeof document === "undefined") return;
+    const observer = new MutationObserver(() => {
+      themeColors = getThemeColors();
+    });
+    observer.observe(document.documentElement, {
+      attributes: true,
+      attributeFilter: ["class"],
+    });
+    return () => observer.disconnect();
+  });
+
  function getNodeLabel(nodeId: string): string {
    const node = data?.nodes?.[nodeId];
    return node?.friendly_name || nodeId.slice(0, 8);
@@ -127,7 +205,7 @@

  function getTemperatureColor(temp: number): string {
    // Default for N/A temp - light gray
-    if (isNaN(temp) || temp === null) return "rgba(179, 179, 179, 0.8)";
+    if (isNaN(temp) || temp === null) return themeColors.wireDefault;

    const coolTemp = 45; // Temp for pure blue
    const midTemp = 57.5; // Temp for pure yellow
@@ -208,7 +286,7 @@
      .append("path")
      .attr("d", "M 0 0 L 10 5 L 0 10")
      .attr("fill", "none")
-      .attr("stroke", "var(--exo-light-gray, #B3B3B3)")
+      .attr("stroke", themeColors.gridStroke)
      .attr("stroke-width", "1.6")
      .attr("stroke-linecap", "round")
      .attr("stroke-linejoin", "round")
@@ -221,7 +299,7 @@
        .attr("y", centerY)
        .attr("text-anchor", "middle")
        .attr("dominant-baseline", "middle")
-        .attr("fill", "rgba(255,215,0,0.4)")
+        .attr("fill", `rgba(${themeColors.accentRgb},0.4)`)
        .attr("font-size", isMinimized ? 10 : 12)
        .attr("font-family", "SF Mono, monospace")
        .attr("letter-spacing", "0.1em")
@@ -505,8 +583,8 @@
              .attr(
                "fill",
                conn.missingIface
-                  ? "rgba(248,113,113,0.9)"
-                  : "rgba(255,255,255,0.85)",
+                  ? themeColors.errorText
+                  : themeColors.normalText,
              )
              .text(label);
            currentY += isTop ? lineHeight : -lineHeight;
@@ -565,22 +643,22 @@

      // Holographic wireframe colors - bright yellow for filter, subtle yellow for hover, grey for filtered out
      const wireColor = isInFilter
-        ? "rgba(255,215,0,1)" // Bright yellow for filter selection
+        ? `rgba(${themeColors.accentRgb},1)` // Bright yellow for filter selection
        : isHovered
-          ? "rgba(255,215,0,0.7)" // Subtle yellow for hover
+          ? `rgba(${themeColors.accentRgb},0.7)` // Subtle yellow for hover
          : isHighlighted
-            ? "rgba(255,215,0,0.9)" // Yellow for instance highlight
+            ? `rgba(${themeColors.accentRgb},0.9)` // Yellow for instance highlight
            : isFilteredOut
-              ? "rgba(140,140,140,0.6)" // Grey for filtered out
-              : "rgba(179,179,179,0.8)"; // Default
-      const wireColorBright = "rgba(255,255,255,0.9)";
+              ? themeColors.wireFiltered // Grey for filtered out
+              : themeColors.wireDefault; // Default
+      const wireColorBright = themeColors.wireBright;
      const fillColor = isInFilter
-        ? "rgba(255,215,0,0.25)"
+        ? `rgba(${themeColors.accentRgb},0.25)`
        : isHovered
-          ? "rgba(255,215,0,0.12)"
+          ? `rgba(${themeColors.accentRgb},0.12)`
          : isHighlighted
-            ? "rgba(255,215,0,0.15)"
-            : "rgba(255,215,0,0.08)";
+            ? `rgba(${themeColors.accentRgb},0.15)`
+            : `rgba(${themeColors.accentRgb},0.08)`;
      const strokeWidth = isInFilter
        ? 3
        : isHovered
@@ -588,8 +666,8 @@
          : isHighlighted
            ? 2.5
            : 1.5;
-      const screenFill = "rgba(0,20,40,0.9)";
-      const glowColor = "rgba(255,215,0,0.3)";
+      const screenFill = themeColors.deviceScreenFill;
+      const glowColor = `rgba(${themeColors.accentRgb},0.3)`;

      const nodeG = nodesGroup
        .append("g")
@@ -653,7 +731,7 @@
          .attr("width", iconBaseWidth)
          .attr("height", iconBaseHeight)
          .attr("rx", cornerRadius)
-          .attr("fill", "#1a1a1a")
+          .attr("fill", themeColors.deviceCase)
          .attr("stroke", wireColor)
          .attr("stroke-width", strokeWidth);

@@ -671,12 +749,12 @@
            )
            .attr("width", iconBaseWidth)
            .attr("height", memFillActualHeight)
-            .attr("fill", "rgba(255,215,0,0.75)")
+            .attr("fill", `rgba(${themeColors.accentRgb},0.75)`)
            .attr("clip-path", `url(#${studioClipId})`);
        }

        // Front panel details - vertical slots
-        const detailColor = "rgba(0,0,0,0.35)";
+        const detailColor = themeColors.detailOverlay;
        const slotHeight = iconBaseHeight * 0.14;
        const vSlotWidth = iconBaseWidth * 0.05;
        const vSlotY =
@@ -736,7 +814,7 @@
          .attr("width", iconBaseWidth)
          .attr("height", iconBaseHeight)
          .attr("rx", cornerRadius)
-          .attr("fill", "#1a1a1a")
+          .attr("fill", themeColors.deviceCase)
          .attr("stroke", wireColor)
          .attr("stroke-width", strokeWidth);

@@ -754,12 +832,12 @@
            )
            .attr("width", iconBaseWidth)
            .attr("height", memFillActualHeight)
-            .attr("fill", "rgba(255,215,0,0.75)")
+            .attr("fill", `rgba(${themeColors.accentRgb},0.75)`)
            .attr("clip-path", `url(#${miniClipId})`);
        }

        // Front panel details - vertical slots (no horizontal slot for Mini)
-        const detailColor = "rgba(0,0,0,0.35)";
+        const detailColor = themeColors.detailOverlay;
        const slotHeight = iconBaseHeight * 0.2;
        const vSlotWidth = iconBaseWidth * 0.045;
        const vSlotY =
@@ -814,7 +892,7 @@
          .attr("width", screenWidth)
          .attr("height", screenHeight)
          .attr("rx", 3)
-          .attr("fill", "#1a1a1a")
+          .attr("fill", themeColors.deviceCase)
          .attr("stroke", wireColor)
          .attr("stroke-width", strokeWidth);

@@ -826,7 +904,7 @@
          .attr("width", screenWidth - screenBezel * 2)
          .attr("height", screenHeight - screenBezel * 2)
          .attr("rx", 2)
-          .attr("fill", "#0a0a12");
+          .attr("fill", themeColors.deviceScreen);

        // Memory fill on screen (fills from bottom up - classic style)
        if (ramUsagePercent > 0) {
@@ -842,7 +920,7 @@
            )
            .attr("width", screenWidth - screenBezel * 2)
            .attr("height", memFillActualHeight)
-            .attr("fill", "rgba(255,215,0,0.85)")
+            .attr("fill", `rgba(${themeColors.accentRgb},0.85)`)
            .attr("clip-path", `url(#${screenClipId})`);
        }

@@ -859,7 +937,7 @@
            "transform",
            `translate(${logoX}, ${logoY}) scale(${logoScale})`,
          )
-          .attr("fill", "#FFFFFF")
+          .attr("fill", themeColors.labelWhite)
          .attr("opacity", 0.9);

        // Base (keyboard) - trapezoidal
@@ -875,7 +953,7 @@
            "d",
            `M ${baseTopX} ${baseY} L ${baseTopX + baseTopWidth} ${baseY} L ${baseBottomX + baseBottomWidth} ${baseY + baseHeight} L ${baseBottomX} ${baseY + baseHeight} Z`,
          )
-          .attr("fill", "#2c2c2c")
+          .attr("fill", themeColors.deviceCaseDark)
          .attr("stroke", wireColor)
          .attr("stroke-width", 1);

@@ -890,7 +968,7 @@
          .attr("y", keyboardY)
          .attr("width", keyboardWidth)
          .attr("height", keyboardHeight)
-          .attr("fill", "rgba(0,0,0,0.2)")
+          .attr("fill", themeColors.deviceShadow)
          .attr("rx", 2);

        // Trackpad
@@ -904,7 +982,7 @@
          .attr("y", trackpadY)
          .attr("width", trackpadWidth)
          .attr("height", trackpadHeight)
-          .attr("fill", "rgba(255,255,255,0.08)")
+          .attr("fill", themeColors.deviceHighlight)
          .attr("rx", 2);
      } else {
        // Default/Unknown - holographic hexagon
@@ -942,7 +1020,7 @@
          .attr("y", gpuBarY)
          .attr("width", gpuBarWidth)
          .attr("height", gpuBarHeight)
-          .attr("fill", "rgba(80, 80, 90, 0.7)")
+          .attr("fill", themeColors.gpuChip)
          .attr("rx", 2);

        // GPU Bar Fill (from bottom up, colored by temperature)
@@ -979,7 +1057,7 @@
          .attr("y", gpuTextY - lineSpacing)
          .attr("text-anchor", "middle")
          .attr("dominant-baseline", "middle")
-          .attr("fill", "#FFFFFF")
+          .attr("fill", themeColors.labelWhite)
          .attr("font-size", gpuTextFontSize)
          .attr("font-weight", "700")
          .attr("font-family", "SF Mono, Monaco, monospace")
@@ -992,7 +1070,7 @@
          .attr("y", gpuTextY)
          .attr("text-anchor", "middle")
          .attr("dominant-baseline", "middle")
-          .attr("fill", "#FFFFFF")
+          .attr("fill", themeColors.labelWhite)
          .attr("font-size", gpuTextFontSize)
          .attr("font-weight", "700")
          .attr("font-family", "SF Mono, Monaco, monospace")
@@ -1005,7 +1083,7 @@
          .attr("y", gpuTextY + lineSpacing)
          .attr("text-anchor", "middle")
          .attr("dominant-baseline", "middle")
-          .attr("fill", "#FFFFFF")
+          .attr("fill", themeColors.labelWhite)
          .attr("font-size", gpuTextFontSize)
          .attr("font-weight", "700")
          .attr("font-family", "SF Mono, Monaco, monospace")
@@ -1033,7 +1111,7 @@
          .attr("y", nameY)
          .attr("text-anchor", "middle")
          .attr("dominant-baseline", "middle")
-          .attr("fill", "#FFD700")
+          .attr("fill", themeColors.accent)
          .attr("font-size", fontSize)
          .attr("font-weight", 500)
          .attr("font-family", "SF Mono, Monaco, monospace")
@@ -1050,15 +1128,15 @@
          .attr("font-family", "SF Mono, Monaco, monospace");
        memText
          .append("tspan")
-          .attr("fill", "rgba(255,215,0,0.9)")
+          .attr("fill", `rgba(${themeColors.accentRgb},0.9)`)
          .text(`${formatBytes(ramUsed)}`);
        memText
          .append("tspan")
-          .attr("fill", "rgba(179,179,179,0.9)")
+          .attr("fill", themeColors.labelMuted)
          .text(`/${formatBytes(ramTotal)}`);
        memText
          .append("tspan")
-          .attr("fill", "rgba(179,179,179,0.7)")
+          .attr("fill", themeColors.labelDim)
          .text(` (${ramUsagePercent.toFixed(0)}%)`);
      } else if (showCompactLabels) {
        // COMPACT MODE: Just name and basic info (4+ nodes)
@@ -1075,7 +1153,7 @@
          .attr("x", nodeInfo.x)
          .attr("y", nameY)
          .attr("text-anchor", "middle")
-          .attr("fill", "#FFD700")
+          .attr("fill", themeColors.accent)
          .attr("font-size", fontSize)
          .attr("font-family", "SF Mono, Monaco, monospace")
          .text(shortName);
@@ -1087,7 +1165,7 @@
          .attr("x", nodeInfo.x)
          .attr("y", statsY)
          .attr("text-anchor", "middle")
-          .attr("fill", "rgba(255,215,0,0.7)")
+          .attr("fill", `rgba(${themeColors.accentRgb},0.7)`)
          .attr("font-size", fontSize * 0.85)
          .attr("font-family", "SF Mono, Monaco, monospace")
          .text(
@@ -1108,7 +1186,7 @@
          .attr("x", nodeInfo.x)
          .attr("y", nameY)
          .attr("text-anchor", "middle")
-          .attr("fill", "#FFD700")
+          .attr("fill", themeColors.accent)
          .attr("font-size", fontSize)
          .attr("font-weight", "500")
          .attr("font-family", "SF Mono, Monaco, monospace")
@@ -1125,15 +1203,15 @@
          .attr("font-family", "SF Mono, Monaco, monospace");
        memTextMini
          .append("tspan")
-          .attr("fill", "rgba(255,215,0,0.9)")
+          .attr("fill", `rgba(${themeColors.accentRgb},0.9)`)
          .text(`${formatBytes(ramUsed)}`);
        memTextMini
          .append("tspan")
-          .attr("fill", "rgba(179,179,179,0.9)")
+          .attr("fill", themeColors.labelMuted)
          .text(`/${formatBytes(ramTotal)}`);
        memTextMini
          .append("tspan")
-          .attr("fill", "rgba(179,179,179,0.7)")
+          .attr("fill", themeColors.labelDim)
          .text(` (${ramUsagePercent.toFixed(0)}%)`);
      }

@@ -1149,8 +1227,8 @@
        const tbStatus = tbBridgeData[nodeInfo.id];
        if (tbStatus) {
          const tbColor = tbStatus.enabled
-            ? "rgba(234,179,8,0.9)"
-            : "rgba(100,100,100,0.7)";
+            ? themeColors.tbActive
+            : themeColors.tbInactive;
          const tbText = tbStatus.enabled ? "TB:ON" : "TB:OFF";
          nodeG
            .append("text")
@@ -1168,7 +1246,7 @@
        if (rdmaStatus !== undefined) {
          const rdmaColor = rdmaStatus.enabled
            ? "rgba(74,222,128,0.9)"
-            : "rgba(100,100,100,0.7)";
+            : themeColors.tbInactive;
          const rdmaText = rdmaStatus.enabled ? "RDMA:ON" : "RDMA:OFF";
          nodeG
            .append("text")
@@ -1189,7 +1267,7 @@
            .attr("x", nodeInfo.x)
            .attr("y", debugLabelY)
            .attr("text-anchor", "middle")
-            .attr("fill", "rgba(179,179,179,0.7)")
+            .attr("fill", themeColors.labelDim)
            .attr("font-size", debugFontSize)
            .attr("font-family", "SF Mono, Monaco, monospace")
            .text(
@@ -1206,6 +1284,7 @@
    const _hoveredNodeId = hoveredNodeId;
    const _filteredNodes = filteredNodes;
    const _highlightedNodes = highlightedNodes;
+    const _themeColors = themeColors;
    if (_data) {
      renderGraph();
    }
@@ -306,13 +306,14 @@ const IMAGE_PARAMS_STORAGE_KEY = "exo-image-generation-params";
 export interface ImageGenerationParams {
  // Basic params
  size:
+    | "auto"
    | "512x512"
    | "768x768"
    | "1024x1024"
    | "1024x768"
    | "768x1024"
-    | "1024x1365"
-    | "1365x1024";
+    | "1024x1536"
+    | "1536x1024";
  quality: "low" | "medium" | "high";
  outputFormat: "png" | "jpeg";
  numImages: number;
@@ -336,7 +337,7 @@ export interface EditingImage {
 }

 const DEFAULT_IMAGE_PARAMS: ImageGenerationParams = {
-  size: "1024x1024",
+  size: "auto",
  quality: "medium",
  outputFormat: "png",
  numImages: 1,
@@ -1,9 +1,12 @@
 <script lang="ts">
  import "../app.css";
+  import { ModeWatcher } from "mode-watcher";

  let { children } = $props();
 </script>

+<ModeWatcher defaultMode="dark" />
+
 <svelte:head>
  <title>EXO</title>
  <meta name="description" content="EXO - Distributed AI Cluster Dashboard" />
@@ -2299,7 +2299,7 @@
              <!-- Panel Header -->
              <div class="flex items-center gap-2 mb-4">
                <div
-                  class="w-2 h-2 bg-exo-yellow rounded-full shadow-[0_0_8px_rgba(255,215,0,0.6)] animate-pulse"
+                  class="w-2 h-2 bg-exo-yellow rounded-full shadow-[0_0_8px_var(--exo-glow-yellow)] animate-pulse"
                ></div>
                <h3
                  class="text-xs text-exo-yellow font-mono tracking-[0.2em] uppercase"
@@ -2948,7 +2948,7 @@
                      <!-- Dot -->
                      <span
                        class="rounded-full transition-all {isSelected
-                          ? 'w-6 h-6 bg-exo-yellow shadow-[0_0_10px_rgba(255,215,0,0.6)]'
+                          ? 'w-6 h-6 bg-exo-yellow shadow-[0_0_10px_var(--exo-glow-yellow)]'
                          : isValid
                            ? 'w-4 h-4 bg-exo-light-gray/70 mt-1'
                            : 'w-3 h-3 bg-exo-medium-gray/50 mt-1.5'}"
@@ -3114,7 +3114,7 @@
                <!-- Panel Header -->
                <div class="flex items-center gap-2 mb-4">
                  <div
-                    class="w-2 h-2 bg-exo-yellow rounded-full shadow-[0_0_8px_rgba(255,215,0,0.6)] animate-pulse"
+                    class="w-2 h-2 bg-exo-yellow rounded-full shadow-[0_0_8px_var(--exo-glow-yellow)] animate-pulse"
                  ></div>
                  <h3
                    class="text-xs text-exo-yellow font-mono tracking-[0.2em] uppercase"
@@ -382,7 +382,7 @@
                class="group border-b border-exo-medium-gray/20 hover:bg-exo-medium-gray/10 transition-colors"
              >
                <td
-                  class="sticky left-0 z-10 bg-exo-dark-gray group-hover:bg-[oklch(0.18_0_0)] transition-colors px-4 py-3 whitespace-nowrap border-r border-exo-medium-gray/20"
+                  class="sticky left-0 z-10 bg-exo-dark-gray group-hover:bg-[var(--exo-bg-hover)] transition-colors px-4 py-3 whitespace-nowrap border-r border-exo-medium-gray/20"
                >
                  <div class="flex items-center gap-2">
                    <div class="min-w-0">
@@ -41,7 +41,7 @@ let

  mlx = stdenv.mkDerivation rec {
    pname = "mlx";
-    version = let v = "0.30.7.dev20260217+50487b41"; in
+    version = let v = "0.30.7.dev20260218+14841977"; 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;
@@ -49,8 +49,8 @@ let
    src = fetchFromGitHub {
      owner = "rltakashige";
      repo = "mlx-jaccl-fix-small-recv";
-      rev = "50487b4141f3c951122655db3b83df5146c1fbeb";
-      hash = "sha256-IL4a9vMX5nocgJU1WG4zE8hArHkHJtnh4sdYh3od5zU=";
+      rev = "1484197707f35186ad3bd614357c7c47fdf86ebc";
+      hash = "sha256-FupCMoK/SF/ldfKuvMSAKECcOP8c+ANgkQlPZttDsLk=";
    };

    patches = [
@@ -19,7 +19,7 @@ dependencies = [
    "anyio==4.11.0",
    "mlx; sys_platform == 'darwin'",
    "mlx[cpu]==0.30.6; sys_platform == 'linux'",
-    "mlx-lm==0.30.6",
+    "mlx-lm==0.30.7",
    "tiktoken>=0.12.0", # required for kimi k2 tokenizer
    "hypercorn>=0.18.0",
    "openai-harmony>=0.0.8",
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "deepseek"
 quantization = "4bit"
 base_model = "DeepSeek V3.1"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 405874409472
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "deepseek"
 quantization = "8bit"
 base_model = "DeepSeek V3.1"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 765577920512
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "glm"
 quantization = "8bit"
 base_model = "GLM 4.5 Air"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 122406567936
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "glm"
 quantization = "bf16"
 base_model = "GLM 4.5 Air"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 229780750336
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "glm"
 quantization = "4bit"
 base_model = "GLM 4.7"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 198556925568
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "glm"
 quantization = "6bit"
 base_model = "GLM 4.7"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 286737579648
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "glm"
 quantization = "8bit"
 base_model = "GLM 4.7"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 396963397248
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "glm"
 quantization = "4bit"
 base_model = "GLM 4.7 Flash"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 19327352832
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "glm"
 quantization = "5bit"
 base_model = "GLM 4.7 Flash"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 22548578304
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "glm"
 quantization = "6bit"
 base_model = "GLM 4.7 Flash"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 26843545600
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "glm"
 quantization = "8bit"
 base_model = "GLM 4.7 Flash"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 34359738368
@@ -0,0 +1,12 @@
+model_id = "mlx-community/GLM-5-8bit-MXFP8"
+n_layers = 78
+hidden_size = 6144
+supports_tensor = true
+tasks = ["TextGeneration"]
+family = "glm"
+quantization = "8bit"
+base_model = "GLM-5"
+capabilities = ["text", "thinking"]
+
+[storage_size]
+in_bytes = 790517400864
@@ -0,0 +1,12 @@
+model_id = "mlx-community/GLM-5-MXFP4-Q8"
+n_layers = 78
+hidden_size = 6144
+supports_tensor = true
+tasks = ["TextGeneration"]
+family = "glm"
+quantization = "MXFP4-Q8"
+base_model = "GLM-5"
+capabilities = ["text", "thinking"]
+
+[storage_size]
+in_bytes = 405478939008
@@ -0,0 +1,12 @@
+model_id = "mlx-community/GLM-5"
+n_layers = 78
+hidden_size = 6144
+supports_tensor = true
+tasks = ["TextGeneration"]
+family = "glm"
+quantization = "bf16"
+base_model = "GLM-5"
+capabilities = ["text", "thinking"]
+
+[storage_size]
+in_bytes = 1487822475264
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "kimi"
 quantization = ""
 base_model = "Kimi K2"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 706522120192
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "kimi"
 quantization = ""
 base_model = "Kimi K2.5"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 662498705408
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "minimax"
 quantization = "3bit"
 base_model = "MiniMax M2.1"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 100086644736
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "minimax"
 quantization = "8bit"
 base_model = "MiniMax M2.1"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 242986745856
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "qwen"
 quantization = "4bit"
 base_model = "Qwen3 0.6B"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 342884352
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "qwen"
 quantization = "8bit"
 base_model = "Qwen3 0.6B"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 698351616
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "qwen"
 quantization = "4bit"
 base_model = "Qwen3 235B"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 141733920768
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "qwen"
 quantization = "8bit"
 base_model = "Qwen3 235B"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 268435456000
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "qwen"
 quantization = "4bit"
 base_model = "Qwen3 30B"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 17612931072
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "qwen"
 quantization = "8bit"
 base_model = "Qwen3 30B"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 33279705088
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "qwen"
 quantization = "4bit"
 base_model = "Qwen3 Next 80B"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 47080074240
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "qwen"
 quantization = "8bit"
 base_model = "Qwen3 Next 80B"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 88814387200
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "step"
 quantization = "4bit"
 base_model = "Step 3.5 Flash"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 114572190076
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "step"
 quantization = "6bit"
 base_model = "Step 3.5 Flash"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 159039627774
@@ -6,7 +6,7 @@ tasks = ["TextGeneration"]
 family = "step"
 quantization = "8bit"
 base_model = "Step 3.5 Flash"
-capabilities = ["text", "thinking"]
+capabilities = ["text", "thinking", "thinking_toggle"]

 [storage_size]
 in_bytes = 209082699847
@@ -25,17 +25,17 @@ workspace = true
 networking = { workspace = true }

 # interop
-pyo3 = { version = "0.27.1", features = [
-    # "abi3-py311", # tells pyo3 (and maturin) to build using the stable ABI with minimum Python version 3.11
+pyo3 = { version = "0.27.2", features = [
+    # "abi3-py313", # tells pyo3 (and maturin) to build using the stable ABI with minimum Python version 3.13
    "nightly", # enables better-supported GIL integration
    "experimental-async", # async support in #[pyfunction] & #[pymethods]
    #"experimental-inspect", # inspection of generated binary => easier to automate type-hint generation
    #"py-clone", # adding Clone-ing of `Py<T>` without GIL (may cause panics - remove if panics happen)
-    "multiple-pymethods", # allows multiple #[pymethods] sections per class
+    # "multiple-pymethods", # allows multiple #[pymethods] sections per class

    # integrations with other libraries
-    "arc_lock", "bigdecimal", "either", "hashbrown", "indexmap", "num-bigint", "num-complex", "num-rational",
-    "ordered-float", "rust_decimal", "smallvec",
+    # "arc_lock", "bigdecimal", "either", "hashbrown", "indexmap", "num-bigint", "num-complex", "num-rational",
+    # "ordered-float", "rust_decimal", "smallvec",
    # "anyhow", "chrono", "chrono-local", "chrono-tz", "eyre", "jiff-02", "lock_api", "parking-lot", "time",  "serde",
 ] }
 pyo3-stub-gen = { version = "0.17.2" }
@@ -45,8 +45,6 @@ pyo3-log = "0.13.2"
 # macro dependencies
 extend = { workspace = true }
 delegate = { workspace = true }
-impl-trait-for-tuples = { workspace = true }
-derive_more = { workspace = true }
 pin-project = { workspace = true }

 # async runtime
@@ -54,24 +52,11 @@ tokio = { workspace = true, features = ["full", "tracing"] }
 futures = { workspace = true }

 # utility dependencies
-once_cell = "1.21.3"
-thread_local = "1.1.9"
 util = { workspace = true }
-thiserror = { workspace = true }
-#internment = { workspace = true }
-#recursion = { workspace = true }
-#generativity = { workspace = true }
-#itertools = { workspace = true }
-

 # Tracing
-#tracing = "0.1"
-#tracing-subscriber = "0.3"
-#console-subscriber = "0.1.5"
-#tracing-log = "0.2.0"
 log = { workspace = true }
 env_logger = "0.11"

-
 # Networking
 libp2p = { workspace = true, features = ["full"] }
@@ -6,7 +6,7 @@ use pyo3::marker::Ungil;
 use pyo3::prelude::*;
 use std::{
    future::Future,
-    pin::{Pin, pin},
+    pin::Pin,
    task::{Context, Poll},
 };

@@ -33,8 +33,6 @@ where

    fn poll(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Self::Output> {
        let waker = cx.waker();
-        Python::with_gil(|py| {
-            py.allow_threads(|| self.project().0.poll(&mut Context::from_waker(waker)))
-        })
+        Python::attach(|py| py.detach(|| self.project().0.poll(&mut Context::from_waker(waker))))
    }
 }
@@ -1,240 +0,0 @@
-//! This module exists to hold examples of some pyo3 patterns that may be too complex to
-//! re-create from scratch, but too inhomogenous to create an abstraction/wrapper around.
-//!
-//! Pattern examples include:
-//!  - Async task handles: with GC-integrated cleanup
-//!  - Sync/async callbacks from python: with propper eventloop handling
-//!
-//! Mutability pattern: https://pyo3.rs/v0.26.0/async-await.html#send--static-constraint
-//!  - Store mutable fields in tokio's `Mutex<T>`
-//!  - For async code: take `&self` and `.lock().await`
-//!  - For sync code: take `&mut self` and `.get_mut()`
-
-use crate::ext::{PyResultExt as _, ResultExt as _, TokioRuntimeExt as _};
-use futures::FutureExt as _;
-use futures::future::BoxFuture;
-use pyo3::exceptions::PyRuntimeError;
-use pyo3::prelude::{PyModule, PyModuleMethods as _};
-use pyo3::{
-    Bound, Py, PyAny, PyErr, PyResult, PyTraverseError, PyVisit, Python, pyclass, pymethods,
-};
-use std::time::Duration;
-use tokio::sync::mpsc;
-use tokio::sync::mpsc::error::TryRecvError;
-
-fn needs_tokio_runtime() {
-    tokio::runtime::Handle::current();
-}
-
-type SyncCallback = Box<dyn Fn() + Send + Sync>;
-type AsyncCallback = Box<dyn Fn() -> BoxFuture<'static, ()> + Send + Sync>;
-
-enum AsyncTaskMessage {
-    SyncCallback(SyncCallback),
-    AsyncCallback(AsyncCallback),
-}
-
-async fn async_task(
-    sender: mpsc::UnboundedSender<()>,
-    mut receiver: mpsc::UnboundedReceiver<AsyncTaskMessage>,
-) {
-    log::info!("RUST: async task started");
-
-    // task state
-    let mut interval = tokio::time::interval(Duration::from_secs(1));
-
-    let mut sync_cbs: Vec<SyncCallback> = vec![];
-    let mut async_cbs: Vec<AsyncCallback> = vec![];
-
-    loop {
-        tokio::select! {
-            // handle incoming messages from task-handle
-            message = receiver.recv() => {
-                // handle closed channel by exiting
-                let Some(message) = message else {
-                    log::info!("RUST: channel closed");
-                    break;
-                };
-
-                // dispatch incoming event
-                match message {
-                    AsyncTaskMessage::SyncCallback(cb) => {
-                        sync_cbs.push(cb);
-                    }
-                    AsyncTaskMessage::AsyncCallback(cb) => {
-                        async_cbs.push(cb);
-                    }
-                }
-            }
-
-            // handle all other events
-            _ = interval.tick() => {
-                log::info!("RUST: async task tick");
-
-                // call back all sync callbacks
-                for cb in &sync_cbs {
-                    cb();
-                }
-
-                // call back all async callbacks
-                for cb in &async_cbs {
-                    cb().await;
-                }
-
-                // send event on unbounded channel
-                sender.send(()).expect("handle receiver cannot be closed/dropped");
-            }
-        }
-    }
-
-    log::info!("RUST: async task stopped");
-}
-
-// #[gen_stub_pyclass]
-#[pyclass(name = "AsyncTaskHandle")]
-#[derive(Debug)]
-struct PyAsyncTaskHandle {
-    sender: Option<mpsc::UnboundedSender<AsyncTaskMessage>>,
-    receiver: mpsc::UnboundedReceiver<()>,
-}
-
-#[allow(clippy::expect_used)]
-impl PyAsyncTaskHandle {
-    const fn sender(&self) -> &mpsc::UnboundedSender<AsyncTaskMessage> {
-        self.sender
-            .as_ref()
-            .expect("The sender should only be None after de-initialization.")
-    }
-
-    const fn sender_mut(&mut self) -> &mpsc::UnboundedSender<AsyncTaskMessage> {
-        self.sender
-            .as_mut()
-            .expect("The sender should only be None after de-initialization.")
-    }
-
-    const fn new(
-        sender: mpsc::UnboundedSender<AsyncTaskMessage>,
-        receiver: mpsc::UnboundedReceiver<()>,
-    ) -> Self {
-        Self {
-            sender: Some(sender),
-            receiver,
-        }
-    }
-}
-
-// #[gen_stub_pymethods]
-#[pymethods]
-impl PyAsyncTaskHandle {
-    #[new]
-    fn py_new(py: Python<'_>) -> PyResult<Self> {
-        use pyo3_async_runtimes::tokio::get_runtime;
-
-        // create communication channel TOWARDS our task
-        let (h_sender, t_receiver) = mpsc::unbounded_channel::<AsyncTaskMessage>();
-
-        // create communication channel FROM our task
-        let (t_sender, h_receiver) = mpsc::unbounded_channel::<()>();
-
-        // perform necessary setup within tokio context - or it crashes
-        let () = get_runtime().block_on(async { needs_tokio_runtime() });
-
-        // spawn tokio task with this thread's task-locals - without this, async callbacks on the new threads will not work!!
-        _ = get_runtime().spawn_with_scope(py, async move {
-            async_task(t_sender, t_receiver).await;
-        });
-        Ok(Self::new(h_sender, h_receiver))
-    }
-
-    /// NOTE: exceptions in callbacks are silently ignored until end of execution
-    fn add_sync_callback(
-        &self,
-        // #[gen_stub(override_type(
-        //     type_repr="collections.abc.Callable[[], None]",
-        //     imports=("collections.abc")
-        // ))]
-        callback: Py<PyAny>,
-    ) -> PyResult<()> {
-        // blocking call to async method -> can do non-blocking if needed
-        self.sender()
-            .send(AsyncTaskMessage::SyncCallback(Box::new(move || {
-                _ = Python::with_gil(|py| callback.call0(py).write_unraisable_with(py));
-            })))
-            .pyerr()?;
-        Ok(())
-    }
-
-    /// NOTE: exceptions in callbacks are silently ignored until end of execution
-    fn add_async_callback(
-        &self,
-        // #[gen_stub(override_type(
-        //     type_repr="collections.abc.Callable[[], collections.abc.Awaitable[None]]",
-        //     imports=("collections.abc")
-        // ))]
-        callback: Py<PyAny>,
-    ) -> PyResult<()> {
-        // blocking call to async method -> can do non-blocking if needed
-        self.sender()
-            .send(AsyncTaskMessage::AsyncCallback(Box::new(move || {
-                let c = Python::with_gil(|py| callback.clone_ref(py));
-                async move {
-                    if let Some(f) = Python::with_gil(|py| {
-                        let coroutine = c.call0(py).write_unraisable_with(py)?;
-                        pyo3_async_runtimes::tokio::into_future(coroutine.into_bound(py))
-                            .write_unraisable_with(py)
-                    }) {
-                        _ = f.await.write_unraisable();
-                    }
-                }
-                .boxed()
-            })))
-            .pyerr()?;
-        Ok(())
-    }
-
-    async fn receive_unit(&mut self) -> PyResult<()> {
-        self.receiver
-            .recv()
-            .await
-            .ok_or(PyErr::new::<PyRuntimeError, _>(
-                "cannot receive unit on closed channel",
-            ))
-    }
-
-    fn drain_units(&mut self) -> PyResult<i32> {
-        let mut cnt = 0;
-        loop {
-            match self.receiver.try_recv() {
-                Err(TryRecvError::Disconnected) => {
-                    return Err(PyErr::new::<PyRuntimeError, _>(
-                        "cannot receive unit on closed channel",
-                    ));
-                }
-                Err(TryRecvError::Empty) => return Ok(cnt),
-                Ok(()) => {
-                    cnt += 1;
-                    continue;
-                }
-            }
-        }
-    }
-
-    // #[gen_stub(skip)]
-    const fn __traverse__(&self, _visit: PyVisit<'_>) -> Result<(), PyTraverseError> {
-        Ok(()) // This is needed purely so `__clear__` can work
-    }
-
-    // #[gen_stub(skip)]
-    fn __clear__(&mut self) {
-        // TODO: may or may not need to await a "kill-signal" oneshot channel message,
-        //       to ensure that the networking task is done BEFORE exiting the clear function...
-        //       but this may require GIL?? and it may not be safe to call GIL here??
-        self.sender = None; // Using Option<T> as a trick to force `sender` channel to be dropped
-    }
-}
-
-pub fn examples_submodule(m: &Bound<'_, PyModule>) -> PyResult<()> {
-    m.add_class::<PyAsyncTaskHandle>()?;
-
-    Ok(())
-}
@@ -17,7 +17,6 @@

 extern crate core;
 mod allow_threading;
-mod examples;
 pub(crate) mod networking;
 pub(crate) mod pylibp2p;

@@ -25,7 +24,6 @@ use crate::networking::networking_submodule;
 use crate::pylibp2p::ident::ident_submodule;
 use crate::pylibp2p::multiaddr::multiaddr_submodule;
 use pyo3::prelude::PyModule;
-use pyo3::prelude::*;
 use pyo3::{Bound, PyResult, pyclass, pymodule};
 use pyo3_stub_gen::define_stub_info_gatherer;

@@ -36,14 +34,10 @@ pub(crate) mod r#const {

 /// Namespace for all the type/trait aliases used by this crate.
 pub(crate) mod alias {
-    use std::error::Error;
    use std::marker::Tuple;

    pub trait SendFn<Args: Tuple + Send + 'static, Output> =
        Fn<Args, Output = Output> + Send + 'static;
-
-    pub type AnyError = Box<dyn Error + Send + Sync + 'static>;
-    pub type AnyResult<T> = Result<T, AnyError>;
 }

 /// Namespace for crate-wide extension traits/methods
@@ -51,7 +45,6 @@ pub(crate) mod ext {
    use crate::allow_threading::AllowThreads;
    use extend::ext;
    use pyo3::exceptions::{PyConnectionError, PyRuntimeError};
-    use pyo3::marker::Ungil;
    use pyo3::types::PyBytes;
    use pyo3::{Py, PyErr, PyResult, Python};
    use tokio::runtime::Runtime;
@@ -62,7 +55,7 @@ pub(crate) mod ext {
    #[ext(pub, name = ByteArrayExt)]
    impl [u8] {
        fn pybytes(&self) -> Py<PyBytes> {
-            Python::with_gil(|py| PyBytes::new(py, self).unbind())
+            Python::attach(|py| PyBytes::new(py, self).unbind())
        }
    }

@@ -98,7 +91,7 @@ pub(crate) mod ext {
    #[ext(pub, name = PyResultExt)]
    impl<T> PyResult<T> {
        fn write_unraisable(self) -> Option<T> {
-            Python::with_gil(|py| self.write_unraisable_with(py))
+            Python::attach(|py| self.write_unraisable_with(py))
        }

        fn write_unraisable_with(self, py: Python<'_>) -> Option<T> {
@@ -175,24 +168,6 @@ pub(crate) mod ext {
    }
 }

-pub(crate) mod private {
-    use std::marker::Sized;
-
-    /// Sealed traits support
-    pub trait Sealed {}
-    impl<T: ?Sized> Sealed for T {}
-}
-
-/// A wrapper around [`Py`] that implements [`Clone`] using [`Python::with_gil`].
-#[repr(transparent)]
-pub(crate) struct ClonePy<T>(pub Py<T>);
-
-impl<T> Clone for ClonePy<T> {
-    fn clone(&self) -> Self {
-        Python::with_gil(|py| Self(self.0.clone_ref(py)))
-    }
-}
-
 /// A Python module implemented in Rust. The name of this function must match
 /// the `lib.name` setting in the `Cargo.toml`, else Python will not be able to
 /// import the module.
@@ -11,9 +11,9 @@ use crate::ext::{ResultExt as _, TokioMpscReceiverExt as _, TokioMpscSenderExt a
 use crate::pyclass;
 use crate::pylibp2p::ident::{PyKeypair, PyPeerId};
 use libp2p::futures::StreamExt as _;
+use libp2p::gossipsub;
 use libp2p::gossipsub::{IdentTopic, Message, MessageId, PublishError};
 use libp2p::swarm::SwarmEvent;
-use libp2p::{gossipsub, mdns};
 use networking::discovery;
 use networking::swarm::create_swarm;
 use pyo3::prelude::{PyModule, PyModuleMethods as _};
@@ -25,7 +25,7 @@ use tokio::sync::{Mutex, mpsc, oneshot};

 mod exception {
    use pyo3::types::PyTuple;
-    use pyo3::{PyErrArguments, exceptions::PyException, prelude::*};
+    use pyo3::{exceptions::PyException, prelude::*};
    use pyo3_stub_gen::derive::*;

    #[gen_stub_pyclass]
@@ -155,7 +155,6 @@ async fn networking_task(
 ) {
    use SwarmEvent::*;
    use ToTask::*;
-    use mdns::Event::*;
    use networking::swarm::BehaviourEvent::*;

    log::info!("RUST: networking task started");
@@ -485,7 +484,7 @@ impl PyNetworkingHandle {
        let (tx, rx) = oneshot::channel();

        // send off request to subscribe
-        let data = Python::with_gil(|py| Vec::from(data.as_bytes(py)));
+        let data = Python::attach(|py| Vec::from(data.as_bytes(py)));
        self.to_task_tx()
            .send_py(ToTask::GossipsubPublish {
                topic,
@@ -19,8 +19,6 @@ either = { workspace = true }
 # macro dependencies
 extend = { workspace = true }
 delegate = { workspace = true }
-impl-trait-for-tuples = { workspace = true }
-derive_more = { workspace = true }

 # async
 tokio = { workspace = true, features = ["full"] }
@@ -29,11 +27,6 @@ futures-timer = { workspace = true }

 # utility dependencies
 util = { workspace = true }
-thiserror = { workspace = true }
-#internment = { workspace = true }
-#recursion = { workspace = true }
-#generativity = { workspace = true }
-#itertools = { workspace = true }
 tracing-subscriber = { version = "0.3.19", features = ["default", "env-filter"] }
 keccak-const = { workspace = true }

@@ -41,4 +34,4 @@ keccak-const = { workspace = true }
 log = { workspace = true }

 # networking
-libp2p = { workspace = true, features = ["full"] }
+libp2p = { workspace = true, features = ["full"] }
@@ -24,8 +24,8 @@ use libp2p::{
    swarm::{NetworkBehaviour, SwarmEvent},
    tcp, yamux,
 };
+use std::error::Error;
 use std::time::Duration;
-use std::{error::Error, hash::Hash};
 use tokio::{io, io::AsyncBufReadExt, select};
 use tracing_subscriber::EnvFilter;

@@ -1,5 +1,4 @@
 use crate::ext::MultiaddrExt;
-use crate::keep_alive;
 use delegate::delegate;
 use either::Either;
 use futures::FutureExt;
@@ -1,44 +0,0 @@
-use delegate::delegate;
-use libp2p::swarm::handler::ConnectionEvent;
-use libp2p::swarm::{ConnectionHandlerEvent, SubstreamProtocol, dummy, handler};
-use std::task::{Context, Poll};
-
-/// An implementation of [`ConnectionHandler`] that doesn't handle any protocols, but it keeps
-/// the connection alive.
-#[derive(Clone)]
-#[repr(transparent)]
-pub struct ConnectionHandler(dummy::ConnectionHandler);
-
-impl ConnectionHandler {
-    pub fn new() -> Self {
-        ConnectionHandler(dummy::ConnectionHandler)
-    }
-}
-
-impl handler::ConnectionHandler for ConnectionHandler {
-    // delegate types and implementation mostly to dummy handler
-    type FromBehaviour = <dummy::ConnectionHandler as handler::ConnectionHandler>::FromBehaviour;
-    type ToBehaviour = <dummy::ConnectionHandler as handler::ConnectionHandler>::ToBehaviour;
-    type InboundProtocol =
-        <dummy::ConnectionHandler as handler::ConnectionHandler>::InboundProtocol;
-    type OutboundProtocol =
-        <dummy::ConnectionHandler as handler::ConnectionHandler>::OutboundProtocol;
-    type InboundOpenInfo =
-        <dummy::ConnectionHandler as handler::ConnectionHandler>::InboundOpenInfo;
-    type OutboundOpenInfo =
-        <dummy::ConnectionHandler as handler::ConnectionHandler>::OutboundOpenInfo;
-
-    delegate! {
-        to self.0 {
-            fn listen_protocol(&self) -> SubstreamProtocol<Self::InboundProtocol, Self::InboundOpenInfo>;
-            fn poll(&mut self, cx: &mut Context<'_>) -> Poll<ConnectionHandlerEvent<Self::OutboundProtocol, Self::OutboundOpenInfo, Self::ToBehaviour>>;
-            fn on_behaviour_event(&mut self, event: Self::FromBehaviour);
-            fn on_connection_event(&mut self, event: ConnectionEvent<Self::InboundProtocol, Self::OutboundProtocol, Self::InboundOpenInfo, Self::OutboundOpenInfo>);
-        }
-    }
-
-    // specifically override this to force connection to stay alive
-    fn connection_keep_alive(&self) -> bool {
-        true
-    }
-}
@@ -3,19 +3,7 @@
 //! this is here as a placeholder documentation
 //!
 //!
-
-// enable Rust-unstable features for convenience
-#![feature(trait_alias)]
-// #![feature(stmt_expr_attributes)]
-// #![feature(unboxed_closures)]
-// #![feature(assert_matches)]
-// #![feature(async_fn_in_dyn_trait)]
-// #![feature(async_for_loop)]
-// #![feature(auto_traits)]
-// #![feature(negative_impls)]
-
 pub mod discovery;
-pub mod keep_alive;
 pub mod swarm;

 /// Namespace for all the type/trait aliases used by this crate.
@@ -54,11 +42,3 @@ pub(crate) mod ext {
        }
    }
 }
-
-pub(crate) mod private {
-    #![allow(dead_code)]
-
-    /// Sealed traits support
-    pub trait Sealed {}
-    impl<T: ?Sized> Sealed for T {}
-}
@@ -47,6 +47,7 @@ class DownloadCoordinator:
    download_command_receiver: Receiver[ForwarderDownloadCommand]
    local_event_sender: Sender[ForwarderEvent]
    event_index_counter: Iterator[int]
+    offline: bool = False

    # Local state
    download_status: dict[ModelId, DownloadProgress] = field(default_factory=dict)
@@ -62,6 +63,8 @@ class DownloadCoordinator:

    def __post_init__(self) -> None:
        self.event_sender, self.event_receiver = channel[Event]()
+        if self.offline:
+            self.shard_downloader.set_internet_connection(False)
        self.shard_downloader.on_progress(self._download_progress_callback)

    def _model_dir(self, model_id: ModelId) -> str:
@@ -107,13 +110,17 @@ class DownloadCoordinator:
            self._last_progress_time[model_id] = current_time()

    async def run(self) -> None:
-        logger.info("Starting DownloadCoordinator")
-        self._test_internet_connection()
+        logger.info(
+            f"Starting DownloadCoordinator{' (offline mode)' if self.offline else ''}"
+        )
+        if not self.offline:
+            self._test_internet_connection()
        async with self._tg as tg:
            tg.start_soon(self._command_processor)
            tg.start_soon(self._forward_events)
            tg.start_soon(self._emit_existing_download_progress)
-            tg.start_soon(self._check_internet_connection)
+            if not self.offline:
+                tg.start_soon(self._check_internet_connection)

    def _test_internet_connection(self) -> None:
        try:
@@ -202,6 +209,20 @@ class DownloadCoordinator:
            )
            return

+        if self.offline:
+            logger.warning(
+                f"Offline mode: model {model_id} is not fully available locally, cannot download"
+            )
+            failed = DownloadFailed(
+                shard_metadata=shard,
+                node_id=self.node_id,
+                error_message=f"Model files not found locally in offline mode: {model_id}",
+                model_directory=self._model_dir(model_id),
+            )
+            self.download_status[model_id] = failed
+            await self.event_sender.send(NodeDownloadProgress(download_progress=failed))
+            return
+
        # Start actual download
        self._start_download_task(shard, initial_progress)

@@ -448,12 +448,13 @@ async def download_file_with_retry(
    target_dir: Path,
    on_progress: Callable[[int, int, bool], None] = lambda _, __, ___: None,
    on_connection_lost: Callable[[], None] = lambda: None,
+    skip_internet: bool = False,
 ) -> Path:
    n_attempts = 3
    for attempt in range(n_attempts):
        try:
            return await _download_file(
-                model_id, revision, path, target_dir, on_progress
+                model_id, revision, path, target_dir, on_progress, skip_internet
            )
        except HuggingFaceAuthenticationError:
            raise
@@ -487,10 +488,14 @@ async def _download_file(
    path: str,
    target_dir: Path,
    on_progress: Callable[[int, int, bool], None] = lambda _, __, ___: None,
+    skip_internet: bool = False,
 ) -> Path:
    target_path = target_dir / path

    if await aios.path.exists(target_path):
+        if skip_internet:
+            return target_path
+
        local_size = (await aios.stat(target_path)).st_size

        # Try to verify against remote, but allow offline operation
@@ -510,6 +515,11 @@ async def _download_file(
            )
            return target_path

+    if skip_internet:
+        raise FileNotFoundError(
+            f"File {path} not found locally and cannot download in offline mode"
+        )
+
    await aios.makedirs((target_dir / path).parent, exist_ok=True)
    length, etag = await file_meta(model_id, revision, path)
    remote_hash = etag[:-5] if etag.endswith("-gzip") else etag
@@ -814,6 +824,7 @@ async def download_shard(
                    file, curr_bytes, total_bytes, is_renamed
                ),
                on_connection_lost=on_connection_lost,
+                skip_internet=skip_internet,
            )

    if not skip_download:
@@ -0,0 +1,230 @@
+"""Tests for offline/air-gapped mode."""
+
+from collections.abc import AsyncIterator
+from pathlib import Path
+from unittest.mock import AsyncMock, patch
+
+import aiofiles
+import aiofiles.os as aios
+import pytest
+
+from exo.download.download_utils import (
+    _download_file,  # pyright: ignore[reportPrivateUsage]
+    download_file_with_retry,
+    fetch_file_list_with_cache,
+)
+from exo.shared.types.common import ModelId
+from exo.shared.types.worker.downloads import FileListEntry
+
+
+@pytest.fixture
+def model_id() -> ModelId:
+    return ModelId("test-org/test-model")
+
+
+@pytest.fixture
+async def temp_models_dir(tmp_path: Path) -> AsyncIterator[Path]:
+    models_dir = tmp_path / "models"
+    await aios.makedirs(models_dir, exist_ok=True)
+    with patch("exo.download.download_utils.EXO_MODELS_DIR", models_dir):
+        yield models_dir
+
+
+class TestDownloadFileOffline:
+    """Tests for _download_file with skip_internet=True."""
+
+    async def test_returns_local_file_without_http_verification(
+        self, model_id: ModelId, tmp_path: Path
+    ) -> None:
+        """When skip_internet=True and file exists locally, return it immediately
+        without making any HTTP calls (no file_meta verification)."""
+        target_dir = tmp_path / "downloads"
+        await aios.makedirs(target_dir, exist_ok=True)
+
+        local_file = target_dir / "model.safetensors"
+        async with aiofiles.open(local_file, "wb") as f:
+            await f.write(b"model weights data")
+
+        with patch(
+            "exo.download.download_utils.file_meta",
+            new_callable=AsyncMock,
+        ) as mock_file_meta:
+            result = await _download_file(
+                model_id,
+                "main",
+                "model.safetensors",
+                target_dir,
+                skip_internet=True,
+            )
+
+            assert result == local_file
+            mock_file_meta.assert_not_called()
+
+    async def test_raises_file_not_found_for_missing_file(
+        self, model_id: ModelId, tmp_path: Path
+    ) -> None:
+        """When skip_internet=True and file does NOT exist locally,
+        raise FileNotFoundError instead of attempting download."""
+        target_dir = tmp_path / "downloads"
+        await aios.makedirs(target_dir, exist_ok=True)
+
+        with pytest.raises(FileNotFoundError, match="offline mode"):
+            await _download_file(
+                model_id,
+                "main",
+                "missing_model.safetensors",
+                target_dir,
+                skip_internet=True,
+            )
+
+    async def test_returns_local_file_in_subdirectory(
+        self, model_id: ModelId, tmp_path: Path
+    ) -> None:
+        """When skip_internet=True and file exists in a subdirectory,
+        return it without HTTP calls."""
+        target_dir = tmp_path / "downloads"
+        subdir = target_dir / "transformer"
+        await aios.makedirs(subdir, exist_ok=True)
+
+        local_file = subdir / "diffusion_pytorch_model.safetensors"
+        async with aiofiles.open(local_file, "wb") as f:
+            await f.write(b"weights")
+
+        with patch(
+            "exo.download.download_utils.file_meta",
+            new_callable=AsyncMock,
+        ) as mock_file_meta:
+            result = await _download_file(
+                model_id,
+                "main",
+                "transformer/diffusion_pytorch_model.safetensors",
+                target_dir,
+                skip_internet=True,
+            )
+
+            assert result == local_file
+            mock_file_meta.assert_not_called()
+
+
+class TestDownloadFileWithRetryOffline:
+    """Tests for download_file_with_retry with skip_internet=True."""
+
+    async def test_propagates_skip_internet_to_download_file(
+        self, model_id: ModelId, tmp_path: Path
+    ) -> None:
+        """Verify skip_internet is passed through to _download_file."""
+        target_dir = tmp_path / "downloads"
+        await aios.makedirs(target_dir, exist_ok=True)
+
+        local_file = target_dir / "config.json"
+        async with aiofiles.open(local_file, "wb") as f:
+            await f.write(b'{"model_type": "qwen2"}')
+
+        with patch(
+            "exo.download.download_utils.file_meta",
+            new_callable=AsyncMock,
+        ) as mock_file_meta:
+            result = await download_file_with_retry(
+                model_id,
+                "main",
+                "config.json",
+                target_dir,
+                skip_internet=True,
+            )
+
+            assert result == local_file
+            mock_file_meta.assert_not_called()
+
+    async def test_file_not_found_does_not_retry(
+        self, model_id: ModelId, tmp_path: Path
+    ) -> None:
+        """FileNotFoundError from offline mode should not trigger retries."""
+        target_dir = tmp_path / "downloads"
+        await aios.makedirs(target_dir, exist_ok=True)
+
+        with pytest.raises(FileNotFoundError):
+            await download_file_with_retry(
+                model_id,
+                "main",
+                "nonexistent.safetensors",
+                target_dir,
+                skip_internet=True,
+            )
+
+
+class TestFetchFileListOffline:
+    """Tests for fetch_file_list_with_cache with skip_internet=True."""
+
+    async def test_uses_cached_file_list(
+        self, model_id: ModelId, temp_models_dir: Path
+    ) -> None:
+        """When skip_internet=True and cache file exists, use it without network."""
+        from pydantic import TypeAdapter
+
+        cache_dir = temp_models_dir / "caches" / model_id.normalize()
+        await aios.makedirs(cache_dir, exist_ok=True)
+
+        cached_list = [
+            FileListEntry(type="file", path="model.safetensors", size=1000),
+            FileListEntry(type="file", path="config.json", size=200),
+        ]
+        cache_file = cache_dir / f"{model_id.normalize()}--main--file_list.json"
+        async with aiofiles.open(cache_file, "w") as f:
+            await f.write(
+                TypeAdapter(list[FileListEntry]).dump_json(cached_list).decode()
+            )
+
+        with patch(
+            "exo.download.download_utils.fetch_file_list_with_retry",
+            new_callable=AsyncMock,
+        ) as mock_fetch:
+            result = await fetch_file_list_with_cache(
+                model_id, "main", skip_internet=True
+            )
+
+            assert result == cached_list
+            mock_fetch.assert_not_called()
+
+    async def test_falls_back_to_local_directory_scan(
+        self, model_id: ModelId, temp_models_dir: Path
+    ) -> None:
+        """When skip_internet=True and no cache but local files exist,
+        build file list from local directory."""
+        import json
+
+        model_dir = temp_models_dir / model_id.normalize()
+        await aios.makedirs(model_dir, exist_ok=True)
+
+        async with aiofiles.open(model_dir / "config.json", "w") as f:
+            await f.write('{"model_type": "qwen2"}')
+
+        index_data = {
+            "metadata": {},
+            "weight_map": {"model.layers.0.weight": "model.safetensors"},
+        }
+        async with aiofiles.open(model_dir / "model.safetensors.index.json", "w") as f:
+            await f.write(json.dumps(index_data))
+
+        async with aiofiles.open(model_dir / "model.safetensors", "wb") as f:
+            await f.write(b"x" * 500)
+
+        with patch(
+            "exo.download.download_utils.fetch_file_list_with_retry",
+            new_callable=AsyncMock,
+        ) as mock_fetch:
+            result = await fetch_file_list_with_cache(
+                model_id, "main", skip_internet=True
+            )
+
+            mock_fetch.assert_not_called()
+            paths = {entry.path for entry in result}
+            assert "config.json" in paths
+            assert "model.safetensors" in paths
+
+    async def test_raises_when_no_cache_and_no_local_files(
+        self, model_id: ModelId, temp_models_dir: Path
+    ) -> None:
+        """When skip_internet=True and neither cache nor local files exist,
+        raise FileNotFoundError."""
+        with pytest.raises(FileNotFoundError, match="No internet"):
+            await fetch_file_list_with_cache(model_id, "main", skip_internet=True)
@@ -39,6 +39,7 @@ class Node:

    node_id: NodeId
    event_index_counter: Iterator[int]
+    offline: bool
    _tg: TaskGroup = field(init=False, default_factory=anyio.create_task_group)

    @classmethod
@@ -68,6 +69,7 @@ class Node:
                download_command_receiver=router.receiver(topics.DOWNLOAD_COMMANDS),
                local_event_sender=router.sender(topics.LOCAL_EVENTS),
                event_index_counter=event_index_counter,
+                offline=args.offline,
            )
        else:
            download_coordinator = None
@@ -132,6 +134,7 @@ class Node:
            api,
            node_id,
            event_index_counter,
+            args.offline,
        )

    async def run(self):
@@ -222,6 +225,7 @@ class Node:
                            ),
                            local_event_sender=self.router.sender(topics.LOCAL_EVENTS),
                            event_index_counter=self.event_index_counter,
+                            offline=self.offline,
                        )
                        self._tg.start_soon(self.download_coordinator.run)
                    if self.worker:
@@ -260,6 +264,9 @@ def main():
    logger.info("Starting EXO")
    logger.info(f"EXO_LIBP2P_NAMESPACE: {os.getenv('EXO_LIBP2P_NAMESPACE')}")

+    if args.offline:
+        logger.info("Running in OFFLINE mode — no internet checks, local models only")
+
    # Set FAST_SYNCH override env var for runner subprocesses
    if args.fast_synch is True:
        os.environ["EXO_FAST_SYNCH"] = "on"
@@ -282,6 +289,7 @@ class Args(CamelCaseModel):
    tb_only: bool = False
    no_worker: bool = False
    no_downloads: bool = False
+    offline: bool = False
    fast_synch: bool | None = None  # None = auto, True = force on, False = force off

    @classmethod
@@ -329,6 +337,11 @@ class Args(CamelCaseModel):
            action="store_true",
            help="Disable the download coordinator (node won't download models)",
        )
+        parser.add_argument(
+            "--offline",
+            action="store_true",
+            help="Run in offline/air-gapped mode: skip internet checks, use only pre-staged local models",
+        )
        fast_synch_group = parser.add_mutually_exclusive_group()
        fast_synch_group.add_argument(
            "--fast-synch",
@@ -85,6 +85,7 @@ from exo.shared.types.api import (
    ImageGenerationTaskParams,
    ImageListItem,
    ImageListResponse,
+    ImageSize,
    ModelList,
    ModelListModel,
    PlaceInstanceParams,
@@ -100,6 +101,7 @@ from exo.shared.types.api import (
    TraceRankStats,
    TraceResponse,
    TraceStatsResponse,
+    normalize_image_size,
 )
 from exo.shared.types.chunks import (
    ErrorChunk,
@@ -751,9 +753,11 @@ class API:
        When stream=True and partial_images > 0, returns a StreamingResponse
        with SSE-formatted events for partial and final images.
        """
-        payload.model = await self._validate_image_model(ModelId(payload.model))
        payload = payload.model_copy(
-            update={"advanced_params": _ensure_seed(payload.advanced_params)}
+            update={
+                "model": await self._validate_image_model(ModelId(payload.model)),
+                "advanced_params": _ensure_seed(payload.advanced_params),
+            }
        )

        command = ImageGeneration(
@@ -1009,12 +1013,13 @@ class API:
    async def bench_image_generations(
        self, request: Request, payload: BenchImageGenerationTaskParams
    ) -> BenchImageGenerationResponse:
-        payload.model = await self._validate_image_model(ModelId(payload.model))
-
-        payload.stream = False
-        payload.partial_images = 0
        payload = payload.model_copy(
-            update={"advanced_params": _ensure_seed(payload.advanced_params)}
+            update={
+                "model": await self._validate_image_model(ModelId(payload.model)),
+                "stream": False,
+                "partial_images": 0,
+                "advanced_params": _ensure_seed(payload.advanced_params),
+            }
        )

        command = ImageGeneration(
@@ -1035,7 +1040,7 @@ class API:
        prompt: str,
        model: ModelId,
        n: int,
-        size: str,
+        size: ImageSize,
        response_format: Literal["url", "b64_json"],
        input_fidelity: Literal["low", "high"],
        stream: bool,
@@ -1105,7 +1110,7 @@ class API:
        prompt: str = Form(...),
        model: str = Form(...),
        n: int = Form(1),
-        size: str = Form("1024x1024"),
+        size: str | None = Form(None),
        response_format: Literal["url", "b64_json"] = Form("b64_json"),
        input_fidelity: Literal["low", "high"] = Form("low"),
        stream: str = Form("false"),
@@ -1131,7 +1136,7 @@ class API:
            prompt=prompt,
            model=ModelId(model),
            n=n,
-            size=size,
+            size=normalize_image_size(size),
            response_format=response_format,
            input_fidelity=input_fidelity,
            stream=stream_bool,
@@ -1167,7 +1172,7 @@ class API:
        prompt: str = Form(...),
        model: str = Form(...),
        n: int = Form(1),
-        size: str = Form("1024x1024"),
+        size: str | None = Form(None),
        response_format: Literal["url", "b64_json"] = Form("b64_json"),
        input_fidelity: Literal["low", "high"] = Form("low"),
        quality: Literal["high", "medium", "low"] = Form("medium"),
@@ -1187,7 +1192,7 @@ class API:
            prompt=prompt,
            model=ModelId(model),
            n=n,
-            size=size,
+            size=normalize_image_size(size),
            response_format=response_format,
            input_fidelity=input_fidelity,
            stream=False,
@@ -44,7 +44,8 @@ async def _refresh_card_cache():
        async for toml_file in path.rglob("*.toml"):
            try:
                card = await ModelCard.load_from_path(toml_file)
-                _card_cache[card.model_id] = card
+                if card.model_id not in _card_cache:
+                    _card_cache[card.model_id] = card
            except (ValidationError, TOMLKitError):
                pass

@@ -182,6 +183,7 @@ class ConfigData(BaseModel):
    def supports_tensor(self) -> bool:
        return self.architectures in [
            ["Glm4MoeLiteForCausalLM"],
+            ["GlmMoeDsaForCausalLM"],
            ["DeepseekV32ForCausalLM"],
            ["DeepseekV3ForCausalLM"],
            ["Qwen3NextForCausalLM"],
@@ -1,9 +1,9 @@
 import time
 from collections.abc import Generator
-from typing import Annotated, Any, Literal
+from typing import Annotated, Any, Literal, get_args
 from uuid import uuid4

-from pydantic import BaseModel, Field
+from pydantic import BaseModel, Field, field_validator

 from exo.shared.models.model_cards import ModelCard, ModelId
 from exo.shared.types.common import CommandId, NodeId
@@ -262,6 +262,27 @@ class DeleteInstanceResponse(BaseModel):
    instance_id: InstanceId


+ImageSize = Literal[
+    "auto",
+    "512x512",
+    "768x768",
+    "1024x768",
+    "768x1024",
+    "1024x1024",
+    "1024x1536",
+    "1536x1024",
+]
+
+
+def normalize_image_size(v: object) -> ImageSize:
+    """Shared validator for ImageSize fields: maps None → "auto" and rejects invalid values."""
+    if v is None:
+        return "auto"
+    if v not in get_args(ImageSize):
+        raise ValueError(f"Invalid size: {v!r}. Must be one of {get_args(ImageSize)}")
+    return v  # pyright: ignore[reportReturnType]
+
+
 class AdvancedImageParams(BaseModel):
    seed: Annotated[int, Field(ge=0)] | None = None
    num_inference_steps: Annotated[int, Field(ge=1, le=100)] | None = None
@@ -281,7 +302,7 @@ class ImageGenerationTaskParams(BaseModel):
    partial_images: int | None = 0
    quality: Literal["high", "medium", "low"] | None = "medium"
    response_format: Literal["url", "b64_json"] | None = "b64_json"
-    size: str | None = "1024x1024"
+    size: ImageSize = "auto"
    stream: bool | None = False
    style: str | None = "vivid"
    user: str | None = None
@@ -289,6 +310,11 @@ class ImageGenerationTaskParams(BaseModel):
    # Internal flag for benchmark mode - set by API, preserved through serialization
    bench: bool = False

+    @field_validator("size", mode="before")
+    @classmethod
+    def normalize_size(cls, v: object) -> ImageSize:
+        return normalize_image_size(v)
+

 class BenchImageGenerationTaskParams(ImageGenerationTaskParams):
    bench: bool = True
@@ -305,13 +331,18 @@ class ImageEditsTaskParams(BaseModel):
    quality: Literal["high", "medium", "low"] | None = "medium"
    output_format: Literal["png", "jpeg", "webp"] = "png"
    response_format: Literal["url", "b64_json"] | None = "b64_json"
-    size: str | None = "1024x1024"
+    size: ImageSize = "auto"
    image_strength: float | None = 0.7
    stream: bool = False
    partial_images: int | None = 0
    advanced_params: AdvancedImageParams | None = None
    bench: bool = False

+    @field_validator("size", mode="before")
+    @classmethod
+    def normalize_size(cls, v: object) -> ImageSize:
+        return normalize_image_size(v)
+
    def __repr_args__(self) -> Generator[tuple[str, Any], None, None]:
        for name, value in super().__repr_args__():  # pyright: ignore[reportAny]
            if name == "image_data":
@@ -14,6 +14,7 @@ from exo.shared.types.api import (
    ImageEditsTaskParams,
    ImageGenerationStats,
    ImageGenerationTaskParams,
+    ImageSize,
 )
 from exo.shared.types.memory import Memory
 from exo.shared.types.worker.runner_response import (
@@ -23,9 +24,9 @@ from exo.shared.types.worker.runner_response import (
 from exo.worker.engines.image.distributed_model import DistributedImageModel


-def parse_size(size_str: str | None) -> tuple[int, int]:
+def parse_size(size_str: ImageSize) -> tuple[int, int]:
    """Parse size parameter like '1024x1024' to (width, height) tuple."""
-    if not size_str:
+    if size_str == "auto":
        return (1024, 1024)

    try:
@@ -109,6 +110,9 @@ def generate_image(
            # Decode base64 image data and save to temp file
            image_path = Path(tmpdir) / "input.png"
            image_path.write_bytes(base64.b64decode(task.image_data))
+            if task.size == "auto":
+                with Image.open(image_path) as img:
+                    width, height = img.size

        for image_num in range(num_images):
            # Increment seed for each image to ensure unique results
@@ -163,11 +163,14 @@ class PipelineLastLayer(CustomMlxLayer):
                output, (self.r + 1) % self.s, group=self.group
            )
            if cache is not None:
-                cache.keys = mx.depends(cache.keys, output)  # type: ignore[reportUnknownMemberType]
+                # CacheList (used by MLA models like DeepSeekV32, GLM MoE DSA)
+                # doesn't have .keys directly; access via first sub-cache.
+                _cache = cache[0] if hasattr(cache, "caches") else cache  # type: ignore
+                _cache.keys = mx.depends(_cache.keys, output)  # type: ignore
            if self.is_prefill:
                mx.eval(output)
                if cache is not None:
-                    mx.eval(cache.keys)  # type: ignore
+                    mx.eval(_cache.keys)  # type: ignore

        if not self.is_prefill:
            output = mx.distributed.all_gather(output, group=self.group)[
@@ -307,7 +310,9 @@ def patch_pipeline_model[T](model: T, group: mx.distributed.Group) -> T:

        # Add dependency to last cache entry to ensure distributed ops are evaluated
        if cache is not None:
-            cache[-1].state = mx.depends(cache[-1].state, logits)  # type: ignore
+            last = cache[-1]  # type: ignore
+            dep_cache = last[0] if hasattr(last, "caches") else last  # type: ignore
+            dep_cache.keys = mx.depends(dep_cache.keys, logits)  # type: ignore

        return logits

@@ -333,7 +338,9 @@ def patch_tensor_model[T](model: T) -> T:

        # Add dependency to last cache entry to ensure distributed ops are evaluated
        if cache is not None and len(cache) > 0:  # pyright: ignore[reportAny]
-            cache[-1].state = mx.depends(cache[-1].state, logits)  # pyright: ignore[reportAny,reportUnknownMemberType]
+            last = cache[-1]  # pyright: ignore[reportAny]
+            dep_cache = last[0] if hasattr(last, "caches") else last  # pyright: ignore[reportAny]
+            dep_cache.keys = mx.depends(dep_cache.keys, logits)  # pyright: ignore[reportAny,reportUnknownMemberType]

        return logits

@@ -547,10 +554,12 @@ class DeepSeekShardingStrategy(TensorParallelShardingStrategy):
        on_timeout: TimeoutCallback | None,
    ) -> nn.Module:
        model = cast(DeepseekV3Model, model)
+
        for layer in model.layers:
            eval_with_timeout(
                layer.parameters(), timeout_seconds / len(model.layers), on_timeout
            )
+
            # Shard the self attention
            if layer.self_attn.q_lora_rank is None:
                layer.self_attn.q_proj = self.all_to_sharded_linear(
@@ -581,12 +590,18 @@ class DeepSeekShardingStrategy(TensorParallelShardingStrategy):
                layer.mlp.down_proj = self.sharded_to_all_linear(layer.mlp.down_proj)
                layer.mlp.up_proj = self.all_to_sharded_linear(layer.mlp.up_proj)

-            # Shard the MoE. Shard in place since the MoE should be responsible
-            # for aggregating the results.
+            # Shard the MoE.
            else:
-                self.all_to_sharded_linear_in_place(layer.mlp.shared_experts.gate_proj)
-                self.sharded_to_all_linear_in_place(layer.mlp.shared_experts.down_proj)
-                self.all_to_sharded_linear_in_place(layer.mlp.shared_experts.up_proj)
+                if getattr(layer.mlp, "shared_experts", None) is not None:
+                    self.all_to_sharded_linear_in_place(
+                        layer.mlp.shared_experts.gate_proj
+                    )
+                    self.sharded_to_all_linear_in_place(
+                        layer.mlp.shared_experts.down_proj
+                    )
+                    self.all_to_sharded_linear_in_place(
+                        layer.mlp.shared_experts.up_proj
+                    )
                self.all_to_sharded_linear_in_place(layer.mlp.switch_mlp.gate_proj)
                self.sharded_to_all_linear_in_place(layer.mlp.switch_mlp.down_proj)
                self.all_to_sharded_linear_in_place(layer.mlp.switch_mlp.up_proj)
@@ -779,8 +794,7 @@ class MiniMaxShardingStrategy(TensorParallelShardingStrategy):

            layer.self_attn = WrappedMiniMaxAttention(layer.self_attn, self.group)  # pyright: ignore[reportAttributeAccessIssue,reportArgumentType]

-            # Shard the MoE. Shard in place since the MoE should be responsible
-            # for aggregating the results.
+            # Shard the MoE.
            self.all_to_sharded_linear_in_place(
                layer.block_sparse_moe.switch_mlp.gate_proj
            )
@@ -893,8 +907,7 @@ class QwenShardingStrategy(TensorParallelShardingStrategy):
                    layer.self_attn.num_attention_heads //= self.N
                    layer.self_attn.num_key_value_heads //= self.N

-            # Shard the MoE. Shard in place since the MoE should be responsible
-            # for aggregating the results.
+            # Shard the MoE.
            if isinstance(layer.mlp, (Qwen3MoeSparseMoeBlock, Qwen3NextSparseMoeBlock)):
                self.all_to_sharded_linear_in_place(layer.mlp.switch_mlp.gate_proj)
                self.sharded_to_all_linear_in_place(layer.mlp.switch_mlp.down_proj)
@@ -57,6 +57,7 @@ def prefill(
    sampler: Callable[[mx.array], mx.array],
    prompt_tokens: mx.array,
    cache: KVCacheType,
+    group: mx.distributed.Group | None = None,
 ) -> tuple[float, int, list[CacheSnapshot]]:
    """Prefill the KV cache with prompt tokens.

@@ -86,6 +87,9 @@ def prefill(

    set_pipeline_prefill(model, is_prefill=True)

+    mx_barrier(group)
+    logger.info("Starting prefill")
+
    # Use max_tokens=1 because max_tokens=0 does not work.
    # We just throw away the generated token - we only care about filling the cache
    for _ in stream_generate(
@@ -305,16 +309,9 @@ def mlx_generate(
    )
    max_stop_len = max((len(s) for s in stop_sequences), default=0)

-    mx_barrier(group)
-    logger.info("Starting prefill")
-
    # Prefill cache with all tokens except the last one
    prefill_tps, prefill_tokens, ssm_snapshots_list = prefill(
-        model,
-        tokenizer,
-        sampler,
-        prompt_tokens[:-1],
-        caches,
+        model, tokenizer, sampler, prompt_tokens[:-1], caches, group
    )
    cache_snapshots: list[CacheSnapshot] | None = ssm_snapshots_list or None

@@ -331,6 +328,7 @@ def mlx_generate(
    think_start = tokenizer.think_start
    think_end = tokenizer.think_end

+    logger.info("Starting decode")
    mx_barrier(group)

    for completion_tokens, out in enumerate(
@@ -285,10 +285,12 @@ def get_eos_token_ids_for_model(model_id: ModelId) -> list[int] | None:
    model_id_lower = model_id.lower()
    if "kimi-k2" in model_id_lower:
        return [163586]
-    elif "glm-4.7-flash" in model_id_lower:
+    elif "glm-5" in model_id_lower or "glm-4.7" in model_id_lower:
+        # For GLM-5 and GLM-4.7
        # 154820: <|endoftext|>, 154827: <|user|>, 154829: <|observation|>
        return [154820, 154827, 154829]
    elif "glm" in model_id_lower:
+        # For GLM-4.5 and older
        return [151336, 151329, 151338]
    return None

@@ -191,7 +191,7 @@ class RunnerSupervisor:
        logger.info("Checking runner's status")
        if self.runner_process.is_alive():
            logger.info("Runner was found to be alive, attempting to join process")
-            await to_thread.run_sync(self.runner_process.join, 1)
+            await to_thread.run_sync(self.runner_process.join, 5)
        rc = self.runner_process.exitcode
        logger.info(f"RunnerSupervisor exited with exit code {rc}")
        if rc == 0:
@@ -378,7 +378,7 @@ dependencies = [
    { name = "loguru", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "mflux", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "mlx", version = "0.30.6", source = { registry = "https://pypi.org/simple" }, extra = ["cpu"], marker = "sys_platform == 'linux'" },
-    { name = "mlx", version = "0.30.7.dev20260217+50487b41", source = { git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git?branch=address-rdma-gpu-locks#50487b4141f3c951122655db3b83df5146c1fbeb" }, marker = "sys_platform == 'darwin'" },
+    { name = "mlx", version = "0.30.7.dev20260218+14841977", source = { git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git?branch=address-rdma-gpu-locks#1484197707f35186ad3bd614357c7c47fdf86ebc" }, marker = "sys_platform == 'darwin'" },
    { name = "mlx-lm", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "msgspec", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "openai-harmony", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
@@ -418,7 +418,7 @@ requires-dist = [
    { name = "mflux", specifier = "==0.15.5" },
    { name = "mlx", marker = "sys_platform == 'darwin'", git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git?branch=address-rdma-gpu-locks" },
    { name = "mlx", extras = ["cpu"], marker = "sys_platform == 'linux'", specifier = "==0.30.6" },
-    { name = "mlx-lm", specifier = "==0.30.6" },
+    { name = "mlx-lm", specifier = "==0.30.7" },
    { name = "msgspec", specifier = ">=0.19.0" },
    { name = "openai-harmony", specifier = ">=0.0.8" },
    { name = "pillow", specifier = ">=11.0,<12.0" },
@@ -1021,7 +1021,7 @@ dependencies = [
    { name = "huggingface-hub", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "matplotlib", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "mlx", version = "0.30.6", source = { registry = "https://pypi.org/simple" }, extra = ["cuda13"], marker = "sys_platform == 'linux'" },
-    { name = "mlx", version = "0.30.7.dev20260217+50487b41", source = { git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git?branch=address-rdma-gpu-locks#50487b4141f3c951122655db3b83df5146c1fbeb" }, marker = "sys_platform == 'darwin'" },
+    { name = "mlx", version = "0.30.7.dev20260218+14841977", source = { git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git?branch=address-rdma-gpu-locks#1484197707f35186ad3bd614357c7c47fdf86ebc" }, marker = "sys_platform == 'darwin'" },
    { name = "numpy", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "opencv-python", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "piexif", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
@@ -1068,8 +1068,8 @@ cuda13 = [

 [[package]]
 name = "mlx"
-version = "0.30.7.dev20260217+50487b41"
-source = { git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git?branch=address-rdma-gpu-locks#50487b4141f3c951122655db3b83df5146c1fbeb" }
+version = "0.30.7.dev20260218+14841977"
+source = { git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git?branch=address-rdma-gpu-locks#1484197707f35186ad3bd614357c7c47fdf86ebc" }
 resolution-markers = [
    "sys_platform == 'darwin'",
 ]
@@ -1100,20 +1100,20 @@ wheels = [

 [[package]]
 name = "mlx-lm"
-version = "0.30.6"
+version = "0.30.7"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
    { name = "jinja2", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
-    { name = "mlx", version = "0.30.7.dev20260217+50487b41", source = { git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git?branch=address-rdma-gpu-locks#50487b4141f3c951122655db3b83df5146c1fbeb" }, marker = "sys_platform == 'darwin'" },
+    { name = "mlx", version = "0.30.7.dev20260218+14841977", source = { git = "https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git?branch=address-rdma-gpu-locks#1484197707f35186ad3bd614357c7c47fdf86ebc" }, marker = "sys_platform == 'darwin'" },
    { name = "numpy", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "protobuf", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "pyyaml", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "sentencepiece", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
    { name = "transformers", marker = "sys_platform == 'darwin' or sys_platform == 'linux'" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/76/cb/815deddc8699b1f694d7e1f9cbed52934c03a8b49432c8add72932bb2f0b/mlx_lm-0.30.6.tar.gz", hash = "sha256:807e042d7040268f1b19190b7eaefd8b2efbff5590a65460974ad4225b91dda1", size = 271733, upload-time = "2026-02-04T21:27:45.741Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/66/0d/56542e2ae13ec6f542d3977d7cff89a205d4f6c5122e0ce23f33265f61c9/mlx_lm-0.30.7.tar.gz", hash = "sha256:e5f31ac58d9f2381f28e1ba639ff903e64f7cff1bdc245c0bc97f72264be329c", size = 275764, upload-time = "2026-02-12T18:41:11.86Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/20/5f/01d281f1fa8a1521d5936659beb4f5ab1f32b463d059263cf9d4cef969d9/mlx_lm-0.30.6-py3-none-any.whl", hash = "sha256:a7405bd581eacc4bf8209d7a6b7f23629585a0d7c6740c2a97e51fee35b3b0e1", size = 379451, upload-time = "2026-02-04T21:27:43.222Z" },
+    { url = "https://files.pythonhosted.org/packages/1e/17/a41c798a3d9cbdc47f39c6db5bba4c2cd199203ead26bf911cb03b644070/mlx_lm-0.30.7-py3-none-any.whl", hash = "sha256:17442a4bf01c4c2d3bca1e647712fe44f19890c3f1eadc8589d389e57b44b9bf", size = 386591, upload-time = "2026-02-12T18:41:10.236Z" },
 ]

 [[package]]
Author	SHA1	Message	Date
Alex Cheema	99b0db0634	chore: apply nix fmt formatting to svelte files Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-02-18 11:46:00 -08:00
Alex Cheema	e3d91dc08a	feat: add light/dark mode toggle to dashboard Wires up the existing mode-watcher dependency (added in #1017 but never used) to provide a one-click light/dark toggle in the header nav. Changes: - app.css: Add full light theme via html:not(.dark) block overriding all --exo-* brand color variables; add utility overrides for text-white, bg-black, scrollbar, command-panel, glow-text, graph links, code blocks, hljs syntax, KaTeX/math, and LaTeX proof/theorem blocks. Fix --exo-glow-yellow to be a color value (not a shadow value) so it works inside Tailwind arbitrary shadow classes. - app.html: Add class="dark" to <html> for FOUC prevention. - +layout.svelte: Add <ModeWatcher defaultMode="dark" />. - HeaderNav.svelte: Add sun/moon toggle button using toggleMode from mode-watcher; use color-mix() for logo drop-shadow. - TopologyGraph.svelte: Add getThemeColors() with MutationObserver on <html> class for reactive D3 color updates. - ModelCard.svelte: Add isDark state + MutationObserver + tc derived color map; replace hardcoded SVG device colors with theme-aware values. - MarkdownContent.svelte: Add scoped light mode overrides for markdown rendering (code blocks, tables, blockquotes, math). - ChatForm, ChatMessages, ChatAttachments: Replace hardcoded rgba shadows with var(--exo-glow-yellow). - ImageParamsPanel: Replace #ffd700 slider thumb colors with var(--exo-yellow). - downloads/+page.svelte: Replace hardcoded hover bg with var(--exo-bg-hover). Dark mode remains the default. Builds with zero errors. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-02-18 11:04:09 -08:00
Alex Cheema	6c322ebb72	feat: only show thinking toggle for models that support it (#1497 ) ## Summary - Adds `thinking_toggle` capability to 26 model cards that support toggling thinking mode on/off - GPT-OSS models (20b, 120b) excluded — they always think and don't support toggling - Dashboard UI updated to check for `thinking_toggle` capability before showing the toggle button ## Test plan - [x] `uv run basedpyright` — 0 errors - [x] `uv run ruff check` — all checks passed - [x] `nix fmt` — 0 files changed - [x] `uv run pytest` — 188 passed, 0 failed - [x] Security review passed (no secrets, eval/exec, innerHTML, or dep changes) 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>	2026-02-18 17:05:00 +00:00
vskiwi	2ebe6216b4	feat: add explicit --offline mode for air-gapped clusters (#1525 ) ## Motivation Closes #1510 There is currently no reliable way to run exo on an air-gapped or offline cluster where models are pre-staged on local disks. The two existing mechanisms — `--no-downloads` and `HF_HUB_OFFLINE=1` — each cover only a subset of the problem: 1. `--no-downloads` blocks model loading: When passed, `DownloadCoordinator` is not created. No `NodeDownloadProgress` events are ever emitted, so `_model_needs_download()` in `plan.py` perpetually returns `DownloadModel`, short-circuiting `_load_model()` and preventing the model from ever being loaded. 2. `HF_HUB_OFFLINE=1` doesn't cover exo's aiohttp code: exo's download pipeline primarily uses raw `aiohttp` for HTTP operations (file list fetching, file downloads, HEAD verification), not the `huggingface_hub` library. These calls will attempt connections and time out on air-gapped networks. 3. `skip_internet` is not propagated to `download_file_with_retry()`: Even when `internet_connection = False`, the `_download_file()` function still makes HTTP HEAD calls via `file_meta()` to verify local files and unconditionally attempts downloads for missing files. ## Changes ### `src/exo/main.py` - Add `--offline` flag to `Args` with env var detection (`EXO_OFFLINE=1`, `HF_HUB_OFFLINE=1`) - Pass `offline` to `DownloadCoordinator` at creation and re-creation (election loop) ### `src/exo/download/coordinator.py` - Add `offline: bool = False` field - In offline mode: set `internet_connection = False` immediately in `__post_init__`, skip `_test_internet_connection()` ping (avoids 3s timeout), skip `_check_internet_connection` periodic loop - In `_start_download()`: if model is not fully available locally, emit `DownloadFailed` with clear message instead of starting a download task ### `src/exo/download/download_utils.py` - Add `skip_internet: bool` parameter to `download_file_with_retry()` and `_download_file()` - When `skip_internet=True` in `_download_file()`: return local file immediately without HTTP HEAD verification; raise `FileNotFoundError` for missing files - Propagate `skip_internet` from `download_shard()` to `download_file_with_retry()` ### `src/exo/download/tests/test_offline_mode.py` (new) - 8 tests covering `_download_file`, `download_file_with_retry`, and `fetch_file_list_with_cache` in offline mode ## Why It Works Unlike `--no-downloads` which disables `DownloadCoordinator` entirely, `--offline` keeps the coordinator running in a restricted mode. The existing `_emit_existing_download_progress()` disk scanner still runs every 60 seconds, emitting `DownloadCompleted` events for pre-staged models. These events flow through the event-sourcing pipeline and populate `state.downloads`, which unblocks `_model_needs_download()` in `plan.py` — no changes to the planning logic required. ``` --offline flag → DownloadCoordinator (offline mode) → Skip 1.1.1.1 ping, internet_connection = False → _emit_existing_download_progress scans disk → Emits DownloadCompleted for pre-staged models → _model_needs_download sees DownloadCompleted → _load_model proceeds normally ``` ## Test Plan ### Automated Testing - `ruff check` — passes - 8 new tests in `test_offline_mode.py` — all pass - 11 existing download tests in `test_download_verification.py` — all pass (no regressions) ### Manual Testing 1. Pre-stage a model on disk (e.g., `~/.exo/models/mlx-community--Qwen3-0.6B-4bit/`) 2. Start exo with `--offline` (or `EXO_OFFLINE=1`) 3. Place an instance via API or dashboard 4. Verify: model loads into memory and inference works without any network calls ### Environment - macOS (Apple Silicon), multi-node cluster with Thunderbolt interconnect - Models pre-staged via rsync / NFS mount	2026-02-18 16:18:09 +00:00
ciaranbor	f54c80b121	Ciaran/image edit api (#1500 ) ## Motivation - Image editing previously ignored input image dimensions, always defaulting to 1024x1024 - Size dropdown was hidden in edit mode, giving users no control over output dimensions - Portrait/landscape presets used non-standard aspect ratios (1024x1365 / 1365x1024) ## Changes - Added "auto" size option that uses input image dimensions for edits, defaults to 1024x1024 for generation - Introduced ImageSize Literal type and normalize_image_size() validator (replaces raw str size fields) - Updated portrait/landscape presets to standard 1024x1536 / 1536x1024 - Made size selector visible in edit mode (previously hidden) - Default size changed from "1024x1024" to "auto" ## Why It Works - "auto" reads actual input image dimensions via PIL at generation time, so edits preserve the original aspect ratio - Pydantic field_validator on both ImageGenerationTaskParams and ImageEditsTaskParams normalizes None → "auto", keeping the API backward-compatible ## Test Plan ### Manual Testing - Verify image edits output at the input image's native resolution when size is "auto" - Verify size dropdown appears and works in both generate and edit modes	2026-02-18 16:05:39 +00:00
rltakashige	48b8f86395	Add support for GLM 5 (#1526 ) ## Motivation Add GLM 5 support in favor of #1513 ## Changes <!-- Describe what you changed in detail --> ## Why It Works <!-- Explain why your approach solves the problem --> ## Test Plan ### Manual Testing <!-- Hardware: (e.g., MacBook Pro M1 Max 32GB, Mac Mini M2 16GB, connected via Thunderbolt 4) --> <!-- What you did: --> <!-- - --> ### Automated Testing <!-- Describe changes to automated tests, or how existing tests cover this change --> <!-- - -->	2026-02-18 14:04:06 +00:00
Evan	5cbd6377a2	prioritize official model cards over custom model cards our old model card search path would override official model cards with custom model cards - our packaged model cards should always be the default here	2026-02-18 13:20:05 +00:00
Evan Quiney	8f01523ddb	remove dead code (#1496 )	2026-02-18 11:43:27 +00:00
Alex Cheema	3addeadea8	Update mlx-lm to 0.30.7 (#1520 ) ## Summary - Bumps `mlx-lm` from 0.30.6 to 0.30.7 in `pyproject.toml` and `uv.lock` ## Test plan - [x] `uv lock` resolves successfully - [x] `basedpyright` — no new errors (63 pre-existing in unrelated `test_tool_call_tracker.py`) - [x] `ruff check` — all checks passed - [x] `nix fmt` — no formatting changes - [x] `pytest` — 188 passed, 1 skipped 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>	2026-02-18 11:14:23 +00:00