reduce logs

previous commits logs were too verbose, this tones them down a bit

AGENTS.md

@@ -91,51 +91,6 @@ From .cursorrules:
 - Catch exceptions only where you can handle them meaningfully
 - Use `@final` and immutability wherever applicable
 
-## API Reference
-
-The API is served at `http://localhost:52415` by default. Key files:
-- `docs/api.md`: Full API documentation
-- `src/exo/master/api.py`: FastAPI implementation
-- `src/exo/shared/types/api.py`: Request/response Pydantic models
-
-### Key Endpoints
-
-```
-GET  /node_id              # Current master node ID
-GET  /state                # Full cluster state (topology, instances, downloads, etc.)
-GET  /events               # Event log for debugging
-
-POST /instance             # Create model instance
-GET  /instance/{id}        # Get instance details
-DELETE /instance/{id}      # Delete instance
-GET  /instance/previews    # Preview placements for a model
-GET  /instance/placement   # Compute placement without creating
-
-GET  /models               # List available models
-GET  /v1/models            # OpenAI-compatible model list
-
-POST /v1/chat/completions  # OpenAI-compatible chat completions (streaming/non-streaming)
-POST /bench/chat/completions # Chat completions with performance stats
-```
-
-### Useful curl Commands
-
-```bash
-# Check cluster state
-curl -s http://localhost:52415/state | python3 -m json.tool
-
-# List models
-curl -s http://localhost:52415/models | python3 -m json.tool
-
-# Preview placements for a model
-curl -s "http://localhost:52415/instance/previews?model_id=llama-3.2-1b" | python3 -m json.tool
-
-# Chat completion
-curl -X POST http://localhost:52415/v1/chat/completions \
-  -H "Content-Type: application/json" \
-  -d '{"model": "llama-3.2-1b", "messages": [{"role": "user", "content": "Hello"}]}'
-```
-
 ## Testing
 
 Tests use pytest-asyncio with `asyncio_mode = "auto"`. Tests are in `tests/` subdirectories alongside the code they test. The `EXO_TESTS=1` env var is set during tests.

app/EXO/EXO/ContentView.swift

@@ -56,6 +56,11 @@ struct ContentView: View {
     }
 
     private var shouldShowLocalNetworkWarning: Bool {
+        // Show warning if local network is not working and EXO is running.
+        // The checker uses a longer timeout on first launch to allow time for
+        // the permission prompt, so this correctly handles both:
+        // 1. User denied permission on first launch
+        // 2. Permission broke after restart (macOS TCC bug)
         if case .notWorking = localNetworkChecker.status {
             return controller.status != .stopped
         }

app/EXO/EXO/Services/LocalNetworkChecker.swift

@@ -5,8 +5,8 @@ import os.log
 /// Checks if the app's local network permission is actually functional.
 ///
 /// macOS local network permission can appear enabled in System Preferences but not
-/// actually work after a restart. This service detects this by creating a UDP
-/// connection to the mDNS multicast address (224.0.0.251:5353).
+/// actually work after a restart. This service uses NWConnection to mDNS multicast
+/// to verify actual connectivity.
 @MainActor
 final class LocalNetworkChecker: ObservableObject {
     enum Status: Equatable {
@@ -35,30 +35,43 @@ final class LocalNetworkChecker: ObservableObject {
     }
 
     private static let logger = Logger(subsystem: "io.exo.EXO", category: "LocalNetworkChecker")
+    private static let hasCompletedInitialCheckKey = "LocalNetworkChecker.hasCompletedInitialCheck"
 
     @Published private(set) var status: Status = .unknown
-    @Published private(set) var lastConnectionState: String = "none"
 
     private var connection: NWConnection?
     private var checkTask: Task<Void, Never>?
 
+    /// Whether we've completed at least one check (stored in UserDefaults)
+    private var hasCompletedInitialCheck: Bool {
+        get { UserDefaults.standard.bool(forKey: Self.hasCompletedInitialCheckKey) }
+        set { UserDefaults.standard.set(newValue, forKey: Self.hasCompletedInitialCheckKey) }
+    }
+
     /// Checks if local network access is working.
     func check() {
         checkTask?.cancel()
         status = .checking
-        lastConnectionState = "connecting"
+
+        // Use longer timeout on first launch to allow time for permission prompt
+        let isFirstCheck = !hasCompletedInitialCheck
+        let timeout: UInt64 = isFirstCheck ? 30_000_000_000 : 3_000_000_000
 
         checkTask = Task { [weak self] in
             guard let self else { return }
-            let result = await self.performCheck()
+
+            Self.logger.info("Checking local network connectivity (first check: \(isFirstCheck))")
+            let result = await self.checkConnectivity(timeout: timeout)
             self.status = result
+            self.hasCompletedInitialCheck = true
+
             Self.logger.info("Local network check complete: \(result.displayText)")
         }
     }
 
-    private func performCheck() async -> Status {
-        Self.logger.info("Checking local network access via UDP multicast")
-
+    /// Checks connectivity using NWConnection to mDNS multicast.
+    /// The connection attempt triggers the permission prompt if not yet shown.
+    private func checkConnectivity(timeout: UInt64) async -> Status {
         connection?.cancel()
         connection = nil
 
@@ -84,22 +97,7 @@ final class LocalNetworkChecker: ObservableObject {
                 continuation.resume(returning: status)
             }
 
-            conn.stateUpdateHandler = { [weak self] state in
-                let stateStr: String
-                switch state {
-                case .setup: stateStr = "setup"
-                case .preparing: stateStr = "preparing"
-                case .ready: stateStr = "ready"
-                case .waiting(let e): stateStr = "waiting(\(e))"
-                case .failed(let e): stateStr = "failed(\(e))"
-                case .cancelled: stateStr = "cancelled"
-                @unknown default: stateStr = "unknown"
-                }
-
-                Task { @MainActor in
-                    self?.lastConnectionState = stateStr
-                }
-
+            conn.stateUpdateHandler = { state in
                 switch state {
                 case .ready:
                     resumeOnce(.working)
@@ -108,6 +106,7 @@ final class LocalNetworkChecker: ObservableObject {
                     if errorStr.contains("54") || errorStr.contains("ECONNRESET") {
                         resumeOnce(.notWorking(reason: "Connection blocked"))
                     }
+                // Otherwise keep waiting - might be showing permission prompt
                 case .failed(let error):
                     let errorStr = "\(error)"
                     if errorStr.contains("65") || errorStr.contains("EHOSTUNREACH")
@@ -127,7 +126,7 @@ final class LocalNetworkChecker: ObservableObject {
             conn.start(queue: .main)
 
             Task {
-                try? await Task.sleep(nanoseconds: 3_000_000_000)
+                try? await Task.sleep(nanoseconds: timeout)
                 let state = conn.state
                 switch state {
                 case .ready:

bench/exo_bench.py

@@ -241,6 +241,9 @@ class PromptSizer:
             ids = tokenizer.apply_chat_template(
                 messages, tokenize=True, add_generation_prompt=True
             )
+            # Fix for transformers 5.x
+            if hasattr(ids, "input_ids"):
+                ids = ids.input_ids
             return int(len(ids))
 
         return count_fn

dashboard/src/lib/components/ChatForm.svelte

@@ -60,12 +60,39 @@
 		return models;
 	});
 
-	// Auto-select the first available model if none is selected
+	// Track previous model IDs to detect newly added models (plain variable to avoid reactive loop)
+	let previousModelIds: Set<string> = new Set();
+
+	// Auto-select the first available model if none is selected, if current selection is stale, or if a new model is added
 	$effect(() => {
 		const models = availableModels();
-		if (models.length > 0 && !currentModel) {
-			setSelectedChatModel(models[0].id);
+		const currentModelIds = new Set(models.map(m => m.id));
+
+		if (models.length > 0) {
+			// Find newly added models (in current but not in previous)
+			const newModels = models.filter(m => !previousModelIds.has(m.id));
+
+			// If no model selected, select the first available
+			if (!currentModel) {
+				setSelectedChatModel(models[0].id);
+			}
+			// If current model is stale (no longer has a running instance), reset to first available
+			else if (!models.some(m => m.id === currentModel)) {
+				setSelectedChatModel(models[0].id);
+			}
+			// If a new model was just added, select it
+			else if (newModels.length > 0 && previousModelIds.size > 0) {
+				setSelectedChatModel(newModels[0].id);
+			}
+		} else {
+			// No instances running - clear the selected model
+			if (currentModel) {
+				setSelectedChatModel('');
+			}
 		}
+
+		// Update previous model IDs for next comparison
+		previousModelIds = currentModelIds;
 	});
 
 	function getInstanceModelId(instanceWrapped: unknown): string {

dashboard/src/routes/+page.svelte

@@ -400,10 +400,8 @@ function toggleInstanceDownloadDetails(nodeId: string): void {
 				const errorText = await response.text();
 				console.error('Failed to launch instance:', errorText);
 			} else {
-				// Auto-select the launched model only if no model is currently selected
-				if (!selectedChatModel()) {
-					setSelectedChatModel(modelId);
-				}
+				// Always auto-select the newly launched model so the user chats to what they just launched
+				setSelectedChatModel(modelId);
 				
 				// Scroll to the bottom of instances container to show the new instance
 				// Use multiple attempts to ensure DOM has updated with the new instance
@@ -763,6 +761,10 @@ function toggleInstanceDownloadDetails(nodeId: string): void {
 	async function deleteInstance(instanceId: string) {
 		if (!confirm(`Delete instance ${instanceId.slice(0, 8)}...?`)) return;
 		
+		// Get the model ID of the instance being deleted before we delete it
+		const deletedInstanceModelId = getInstanceModelId(instanceData[instanceId]);
+		const wasSelected = selectedChatModel() === deletedInstanceModelId;
+		
 		try {
 			const response = await fetch(`/instance/${instanceId}`, {
 				method: 'DELETE',
@@ -771,6 +773,24 @@ function toggleInstanceDownloadDetails(nodeId: string): void {
 			
 			if (!response.ok) {
 				console.error('Failed to delete instance:', response.status);
+			} else if (wasSelected) {
+				// If we deleted the currently selected model, switch to another available model
+				// Find another instance that isn't the one we just deleted
+				const remainingInstances = Object.entries(instanceData).filter(([id]) => id !== instanceId);
+				if (remainingInstances.length > 0) {
+					// Select the last instance (most recently added, since objects preserve insertion order)
+					const [, lastInstance] = remainingInstances[remainingInstances.length - 1];
+					const newModelId = getInstanceModelId(lastInstance);
+					if (newModelId && newModelId !== 'Unknown' && newModelId !== 'Unknown Model') {
+						setSelectedChatModel(newModelId);
+					} else {
+						// Clear selection if no valid model found
+						setSelectedChatModel('');
+					}
+				} else {
+					// No more instances, clear the selection
+					setSelectedChatModel('');
+				}
 			}
 		} catch (error) {
 			console.error('Error deleting instance:', error);

justfile

@@ -1,3 +1,5 @@
+export NIX_CONFIG := "extra-experimental-features = nix-command flakes"
+
 fmt:
     nix fmt
 

pyproject.toml

@@ -23,6 +23,7 @@ dependencies = [
     "tiktoken>=0.12.0", # required for kimi k2 tokenizer
     "hypercorn>=0.18.0",
     "openai-harmony>=0.0.8",
+    "httpx>=0.28.1",
 ]
 
 [project.scripts]

src/exo/master/api.py

@@ -13,12 +13,6 @@ from hypercorn.asyncio import serve  # pyright: ignore[reportUnknownVariableType
 from hypercorn.config import Config
 from hypercorn.typing import ASGIFramework
 from loguru import logger
-from openai_harmony import (  # pyright: ignore[reportMissingTypeStubs]
-    HarmonyEncodingName,
-    Role,
-    StreamableParser,
-    load_harmony_encoding,
-)
 
 from exo.master.placement import place_instance as get_instance_placements
 from exo.shared.apply import apply
@@ -67,8 +61,6 @@ from exo.utils.channels import Receiver, Sender, channel
 from exo.utils.dashboard_path import find_dashboard
 from exo.utils.event_buffer import OrderedBuffer
 
-encoding = load_harmony_encoding(HarmonyEncodingName.HARMONY_GPT_OSS)
-
 
 def chunk_to_response(
     chunk: TokenChunk, command_id: CommandId
@@ -381,35 +373,8 @@ class API:
             instance_id=instance_id,
         )
 
-    async def _process_gpt_oss(self, token_chunks: Receiver[TokenChunk]):
-        stream = StreamableParser(encoding, role=Role.ASSISTANT)
-        thinking = False
-
-        async for chunk in token_chunks:
-            stream.process(chunk.token_id)
-
-            delta = stream.last_content_delta
-            ch = stream.current_channel
-
-            if ch == "analysis" and not thinking:
-                thinking = True
-                yield chunk.model_copy(update={"text": "<think>"})
-
-            if ch != "analysis" and thinking:
-                thinking = False
-                yield chunk.model_copy(update={"text": "</think>"})
-
-            if delta:
-                yield chunk.model_copy(update={"text": delta})
-
-            if chunk.finish_reason is not None:
-                if thinking:
-                    yield chunk.model_copy(update={"text": "</think>"})
-                yield chunk
-                break
-
     async def _chat_chunk_stream(
-        self, command_id: CommandId, parse_gpt_oss: bool
+        self, command_id: CommandId
     ) -> AsyncGenerator[TokenChunk, None]:
         """Yield `TokenChunk`s for a given command until completion."""
 
@@ -417,16 +382,10 @@ class API:
             self._chat_completion_queues[command_id], recv = channel[TokenChunk]()
 
             with recv as token_chunks:
-                if parse_gpt_oss:
-                    async for chunk in self._process_gpt_oss(token_chunks):
-                        yield chunk
-                        if chunk.finish_reason is not None:
-                            break
-                else:
-                    async for chunk in token_chunks:
-                        yield chunk
-                        if chunk.finish_reason is not None:
-                            break
+                async for chunk in token_chunks:
+                    yield chunk
+                    if chunk.finish_reason is not None:
+                        break
 
         except anyio.get_cancelled_exc_class():
             # TODO: TaskCancelled
@@ -442,11 +401,11 @@ class API:
             del self._chat_completion_queues[command_id]
 
     async def _generate_chat_stream(
-        self, command_id: CommandId, parse_gpt_oss: bool
+        self, command_id: CommandId
     ) -> AsyncGenerator[str, None]:
         """Generate chat completion stream as JSON strings."""
 
-        async for chunk in self._chat_chunk_stream(command_id, parse_gpt_oss):
+        async for chunk in self._chat_chunk_stream(command_id):
             chunk_response: ChatCompletionResponse = chunk_to_response(
                 chunk, command_id
             )
@@ -458,7 +417,7 @@ class API:
                 yield "data: [DONE]\n\n"
 
     async def _collect_chat_completion(
-        self, command_id: CommandId, parse_gpt_oss: bool
+        self, command_id: CommandId
     ) -> ChatCompletionResponse:
         """Collect all token chunks for a chat completion and return a single response."""
 
@@ -466,7 +425,7 @@ class API:
         model: str | None = None
         finish_reason: FinishReason | None = None
 
-        async for chunk in self._chat_chunk_stream(command_id, parse_gpt_oss):
+        async for chunk in self._chat_chunk_stream(command_id):
             if model is None:
                 model = chunk.model
 
@@ -495,7 +454,7 @@ class API:
         )
 
     async def _collect_chat_completion_with_stats(
-        self, command_id: CommandId, parse_gpt_oss: bool
+        self, command_id: CommandId
     ) -> BenchChatCompletionResponse:
         text_parts: list[str] = []
         model: str | None = None
@@ -503,7 +462,7 @@ class API:
 
         stats: GenerationStats | None = None
 
-        async for chunk in self._chat_chunk_stream(command_id, parse_gpt_oss):
+        async for chunk in self._chat_chunk_stream(command_id):
             if model is None:
                 model = chunk.model
 
@@ -544,8 +503,6 @@ class API:
         """Handle chat completions, supporting both streaming and non-streaming responses."""
         model_meta = await resolve_model_meta(payload.model)
         payload.model = model_meta.model_id
-        parse_gpt_oss = "gpt-oss" in model_meta.model_id.lower()
-        logger.info(f"{parse_gpt_oss=}")
 
         if not any(
             instance.shard_assignments.model_id == payload.model
@@ -562,17 +519,16 @@ class API:
         await self._send(command)
         if payload.stream:
             return StreamingResponse(
-                self._generate_chat_stream(command.command_id, parse_gpt_oss),
+                self._generate_chat_stream(command.command_id),
                 media_type="text/event-stream",
             )
 
-        return await self._collect_chat_completion(command.command_id, parse_gpt_oss)
+        return await self._collect_chat_completion(command.command_id)
 
     async def bench_chat_completions(
         self, payload: BenchChatCompletionTaskParams
     ) -> BenchChatCompletionResponse:
         model_meta = await resolve_model_meta(payload.model)
-        parse_gpt_oss = "gpt-oss" in model_meta.model_id.lower()
         payload.model = model_meta.model_id
 
         if not any(
@@ -589,10 +545,7 @@ class API:
         command = ChatCompletion(request_params=payload)
         await self._send(command)
 
-        response = await self._collect_chat_completion_with_stats(
-            command.command_id,
-            parse_gpt_oss,
-        )
+        response = await self._collect_chat_completion_with_stats(command.command_id)
         return response
 
     def _calculate_total_available_memory(self) -> Memory:

src/exo/shared/logging.py

@@ -29,6 +29,11 @@ class _InterceptHandler(logging.Handler):
 
 def logger_setup(log_file: Path | None, verbosity: int = 0):
     """Set up logging for this process - formatting, file handles, verbosity and output"""
+
+    logging.getLogger("exo_pyo3_bindings").setLevel(logging.WARNING)
+    logging.getLogger("httpx").setLevel(logging.WARNING)
+    logging.getLogger("httpcore").setLevel(logging.WARNING)
+
     logger.remove()
 
     # replace all stdlib loggers with _InterceptHandlers that log to loguru

src/exo/shared/models/model_cards.py

@@ -425,15 +425,15 @@ MODEL_CARDS: dict[str, ModelCard] = {
             supports_tensor=True,
         ),
     ),
-    "gpt-oss-20b-4bit": ModelCard(
-        short_id="gpt-oss-20b-4bit",
-        model_id=ModelId("mlx-community/gpt-oss-20b-MXFP4-Q4"),
-        name="GPT-OSS 20B (MXFP4-Q4, MLX)",
-        description="""OpenAI's GPT-OSS 20B is a medium-sized MoE model for lower-latency and local or specialized use cases; this MLX variant uses MXFP4 4-bit quantization.""",
+    "gpt-oss-20b-MXFP4-Q8": ModelCard(
+        short_id="gpt-oss-20b-MXFP4-Q8",
+        model_id=ModelId("mlx-community/gpt-oss-20b-MXFP4-Q8"),
+        name="GPT-OSS 20B (MXFP4-Q8, MLX)",
+        description="""OpenAI's GPT-OSS 20B is a medium-sized MoE model for lower-latency and local or specialized use cases; this variant is a 4-bit MLX conversion for Apple Silicon.""",
         tags=[],
         metadata=ModelMetadata(
-            model_id=ModelId("mlx-community/gpt-oss-20b-MXFP4-Q4"),
-            pretty_name="GPT-OSS 20B (MXFP4-Q4, MLX)",
+            model_id=ModelId("mlx-community/gpt-oss-20b-MXFP4-Q8"),
+            pretty_name="GPT-OSS 20B (MXFP4-Q8, MLX)",
             storage_size=Memory.from_kb(11_744_051),
             n_layers=24,
             hidden_size=2880,

src/exo/worker/engines/mlx/auto_parallel.py

@@ -228,10 +228,15 @@ def tensor_auto_parallel(
         group=group,
     )
 
-    logger.info(f"tensor_auto_parallel: model type = {type(model).__name__}")
+    if hasattr(model, "shard"):
+        try:
+            model.shard(group)  # type: ignore
+            return model
+        except (AttributeError, TypeError, NameError):
+            pass
 
     if isinstance(model, (LlamaModel, Ministral3Model)):
-        logger.info("Using LlamaShardingStrategy")
+        logger.warning("shouldn't be hit - upstream sharding exists")
         tensor_parallel_sharding_strategy = LlamaShardingStrategy(
             group,
             all_to_sharded_linear,
@@ -240,7 +245,7 @@ def tensor_auto_parallel(
             sharded_to_all_linear_in_place,
         )
     elif isinstance(model, (DeepseekV3Model, DeepseekV32Model)):
-        logger.info("Using DeepSeekShardingStrategy")
+        logger.warning("shouldn't be hit - upstream sharding exists")
         tensor_parallel_sharding_strategy = DeepSeekShardingStrategy(
             group,
             all_to_sharded_linear,
@@ -249,7 +254,6 @@ def tensor_auto_parallel(
             sharded_to_all_linear_in_place,
         )
     elif isinstance(model, MiniMaxModel):
-        logger.info("Using MiniMaxShardingStrategy")
         tensor_parallel_sharding_strategy = MiniMaxShardingStrategy(
             group,
             all_to_sharded_linear,
@@ -258,7 +262,6 @@ def tensor_auto_parallel(
             sharded_to_all_linear_in_place,
         )
     elif isinstance(model, (Qwen3MoeModel, Glm4MoeModel, Qwen3NextModel)):
-        logger.info("Using QwenShardingStrategy")
         tensor_parallel_sharding_strategy = QwenShardingStrategy(
             group,
             all_to_sharded_linear,
@@ -267,7 +270,6 @@ def tensor_auto_parallel(
             sharded_to_all_linear_in_place,
         )
     elif isinstance(model, GptOssModel):
-        logger.info("Using GptOssShardingStrategy for tensor parallelism")
         tensor_parallel_sharding_strategy = GptOssShardingStrategy(
             group,
             all_to_sharded_linear,
@@ -350,8 +352,6 @@ def _set_layers(model: nn.Module, layers: list[_LayerCallable]) -> None:
 class DeepSeekShardingStrategy(TensorParallelShardingStrategy):
     def shard_model(self, model: nn.Module) -> nn.Module:
         model = cast(DeepseekV3Model, model)
-        dense_count = 0
-        moe_count = 0
         for layer in model.layers:
             # Shard the self attention
             if layer.self_attn.q_lora_rank is None:
@@ -370,7 +370,6 @@ class DeepSeekShardingStrategy(TensorParallelShardingStrategy):
 
             # Shard the MLP
             if isinstance(layer.mlp, (DeepseekV3MLP, DeepseekV32MLP)):
-                dense_count += 1
                 layer.mlp.gate_proj = self.all_to_sharded_linear(layer.mlp.gate_proj)
                 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)
@@ -378,7 +377,6 @@ class DeepSeekShardingStrategy(TensorParallelShardingStrategy):
             # Shard the MoE. Shard in place since the MoE should be responsible
             # for aggregating the results.
             else:
-                moe_count += 1
                 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)
@@ -388,7 +386,6 @@ class DeepSeekShardingStrategy(TensorParallelShardingStrategy):
                 layer.mlp = ShardedDeepseekV3MoE(layer.mlp)  # type: ignore
                 layer.mlp.sharding_group = self.group
 
-        logger.info(f"DeepSeekShardingStrategy: {dense_count} dense layers (shard_linear), {moe_count} MoE layers (shard_inplace)")
         return model
 
 
@@ -484,6 +481,7 @@ class ShardedQwenMoE(CustomMlxLayer):
 class GptOssShardingStrategy(TensorParallelShardingStrategy):
     def shard_model(self, model: nn.Module) -> nn.Module:
         model = cast(GptOssMoeModel, model)
+
         for layer in model.layers:
             layer.self_attn.q_proj = self.all_to_sharded_linear(layer.self_attn.q_proj)
             layer.self_attn.k_proj = self.all_to_sharded_linear(layer.self_attn.k_proj)

src/exo/worker/engines/mlx/utils_mlx.py

@@ -20,6 +20,7 @@ except ImportError:
 
 from mlx_lm.models.cache import KVCache, QuantizedKVCache, RotatingKVCache
 from mlx_lm.models.deepseek_v3 import DeepseekV3Model
+from mlx_lm.models.gpt_oss import Model as GptOssModel
 from mlx_lm.tokenizer_utils import TokenizerWrapper
 
 from exo.worker.engines.mlx.constants import (
@@ -162,9 +163,7 @@ def mlx_distributed_init(
                 os.environ["MLX_IBV_DEVICES"] = coordination_file
                 os.environ["MLX_RANK"] = str(rank)
                 os.environ["MLX_JACCL_COORDINATOR"] = jaccl_coordinator
-                logger.info(f"rank {rank} BEFORE mx.distributed.init(backend='jaccl')")
                 group = mx.distributed.init(backend="jaccl", strict=True)
-                logger.info(f"rank {rank} AFTER mx.distributed.init - group created")
 
         logger.info(f"Rank {rank} mlx distributed initialization complete")
 
@@ -201,12 +200,10 @@ def load_mlx_items(
         tokenizer = get_tokenizer(model_path, bound_instance.bound_shard)
 
     else:
-        logger.info("Starting distributed shard_and_load")
+        logger.info("Starting distributed init")
         start_time = time.perf_counter()
-        logger.info(f"BEFORE shard_and_load for model {bound_instance.bound_shard.model_meta.model_id}")
         model, tokenizer = shard_and_load(bound_instance.bound_shard, group=group)
         end_time = time.perf_counter()
-        logger.info(f"AFTER shard_and_load completed")
         logger.info(
             f"Time taken to shard and load model: {(end_time - start_time):.2f}s"
         )
@@ -221,10 +218,8 @@ def shard_and_load(
     group: Group,
 ) -> tuple[nn.Module, TokenizerWrapper]:
     model_path = build_model_path(shard_metadata.model_meta.model_id)
-    logger.info(f"shard_and_load: model_path={model_path}")
-    logger.info("BEFORE load_model (lazy=True)")
+
     model, _ = load_model(model_path, lazy=True, strict=False)
-    logger.info("AFTER load_model")
     logger.debug(model)
     if hasattr(model, "model") and isinstance(model.model, DeepseekV3Model):  # type: ignore
         pass
@@ -258,6 +253,8 @@ def shard_and_load(
             model = pipeline_auto_parallel(model, group, shard_metadata)
 
     mx.eval(model.parameters())
+
+    # TODO: Do we need this?
     mx.eval(model)
 
     logger.debug("SHARDED")
@@ -369,6 +366,8 @@ def apply_chat_template(
         tools=chat_task_data.tools,
     )
 
+    logger.info(prompt)
+
     return prompt
 
 
@@ -400,6 +399,11 @@ def make_kv_cache(
 ) -> list[KVCache | RotatingKVCache | QuantizedKVCache]:
     assert hasattr(model, "layers")
 
+    # TODO: Do this for all models
+    if hasattr(model, "make_cache") and isinstance(model, GptOssModel):
+        logger.info("Using MLX LM's make cache")
+        return model.make_cache()  # type: ignore
+
     if max_kv_size is None:
         if KV_CACHE_BITS is None:
             logger.info("Using default KV cache")

src/exo/worker/runner/runner.py

@@ -1,6 +1,15 @@
 import time
+from collections.abc import Generator
+from functools import cache
 
 import mlx.core as mx
+from mlx_lm.models.gpt_oss import Model as GptOssModel
+from openai_harmony import (  # pyright: ignore[reportMissingTypeStubs]
+    HarmonyEncodingName,
+    Role,
+    StreamableParser,
+    load_harmony_encoding,
+)
 
 from exo.shared.types.api import ChatCompletionMessageText
 from exo.shared.types.chunks import TokenChunk
@@ -153,11 +162,19 @@ def main(
                     _check_for_debug_prompts(task_params.messages[0].content)
 
                     # Generate responses using the actual MLX generation
-                    for response in mlx_generate(
+                    mlx_generator = mlx_generate(
                         model=model,
                         tokenizer=tokenizer,
                         task=task_params,
-                    ):
+                    )
+
+                    # GPT-OSS specific parsing to match other model formats.
+                    if isinstance(model, GptOssModel):
+                        mlx_generator = parse_gpt_oss(mlx_generator)
+
+                    # TODO: Add tool call parser here
+
+                    for response in mlx_generator:
                         match response:
                             case GenerationResponse():
                                 if shard_metadata.device_rank == 0:
@@ -207,6 +224,43 @@ def main(
                 break
 
 
+@cache
+def get_gpt_oss_encoding():
+    encoding = load_harmony_encoding(HarmonyEncodingName.HARMONY_GPT_OSS)
+    return encoding
+
+
+def parse_gpt_oss(
+    responses: Generator[GenerationResponse],
+) -> Generator[GenerationResponse]:
+    encoding = get_gpt_oss_encoding()
+    stream = StreamableParser(encoding, role=Role.ASSISTANT)
+    thinking = False
+
+    for response in responses:
+        stream.process(response.token)
+
+        delta = stream.last_content_delta
+        ch = stream.current_channel
+
+        if ch == "analysis" and not thinking:
+            thinking = True
+            yield response.model_copy(update={"text": "<think>"})
+
+        if ch != "analysis" and thinking:
+            thinking = False
+            yield response.model_copy(update={"text": "</think>"})
+
+        if delta:
+            yield response.model_copy(update={"text": delta})
+
+        if response.finish_reason is not None:
+            if thinking:
+                yield response.model_copy(update={"text": "</think>"})
+            yield response
+            break
+
+
 EXO_RUNNER_MUST_FAIL = "EXO RUNNER MUST FAIL"
 EXO_RUNNER_MUST_OOM = "EXO RUNNER MUST OOM"
 EXO_RUNNER_MUST_TIMEOUT = "EXO RUNNER MUST TIMEOUT"

src/exo/worker/utils/net_profile.py

@@ -1,62 +1,64 @@
-import http.client
-import time
-
-from anyio import create_task_group, to_thread
+import anyio
+import httpx
+from anyio import create_task_group
 from loguru import logger
 
 from exo.shared.topology import Topology
 from exo.shared.types.common import NodeId
 
-BAD_STATUSLINE_ATTEMPTS = 3
+REACHABILITY_ATTEMPTS = 3
 
 
 async def check_reachability(
     target_ip: str,
     expected_node_id: NodeId,
-    self_node_id: NodeId,
     out: dict[NodeId, set[str]],
+    client: httpx.AsyncClient,
 ) -> None:
     """Check if a node is reachable at the given IP and verify its identity."""
+    if ":" in target_ip:
+        # TODO: use real IpAddress types
+        target_ip = f"[{target_ip}]"
+    url = f"http://{target_ip}:52415/node_id"
 
-    # TODO: use an async http client
-    def _fetch_remote_node_id(*, attempt: int = 1) -> NodeId | None:
-        connection = http.client.HTTPConnection(target_ip, 52415, timeout=3)
+    remote_node_id = None
+    last_error = None
+
+    for _ in range(REACHABILITY_ATTEMPTS):
         try:
-            connection.request("GET", "/node_id")
-            response = connection.getresponse()
-            if response.status != 200:
-                return None
+            r = await client.get(url)
+            if r.status_code != 200:
+                await anyio.sleep(1)
+                continue
 
-            body = response.read().decode("utf-8").strip()
+            body = r.text.strip().strip('"')
+            if not body:
+                await anyio.sleep(1)
+                continue
 
-            # Strip quotes if present (JSON string response)
-            if body.startswith('"') and body.endswith('"') and len(body) >= 2:
-                body = body[1:-1]
+            remote_node_id = NodeId(body)
+            break
 
-            return NodeId(body) or None
-        except OSError:
-            return None
-        except http.client.BadStatusLine:
-            if attempt >= BAD_STATUSLINE_ATTEMPTS:
-                logger.warning(
-                    f"BadStatusLine from {target_ip}, after {attempt} attempts, assuming connection to {expected_node_id} has dropped"
-                )
-                return None
-            time.sleep(1)
-            return _fetch_remote_node_id(attempt=attempt + 1)
-        except http.client.HTTPException as e:
-            logger.warning(f"HTTPException from {target_ip}: {type(e).__name__}: {e}")
-            return None
-        finally:
-            connection.close()
+        # expected failure cases
+        except (
+            httpx.TimeoutException,
+            httpx.NetworkError,
+        ):
+            await anyio.sleep(1)
+
+        # other failures should be logged on last attempt
+        except httpx.HTTPError as e:
+            last_error = e
+            await anyio.sleep(1)
+
+    if last_error is not None:
+        logger.warning(
+            f"connect error {type(last_error).__name__} from {target_ip} after {REACHABILITY_ATTEMPTS} attempts; treating as down"
+        )
 
-    remote_node_id = await to_thread.run_sync(_fetch_remote_node_id)
     if remote_node_id is None:
         return
 
-    if remote_node_id == self_node_id:
-        return
-
     if remote_node_id != expected_node_id:
         logger.warning(
             f"Discovered node with unexpected node_id; "
@@ -74,18 +76,33 @@ async def check_reachable(
     topology: Topology, self_node_id: NodeId
 ) -> dict[NodeId, set[str]]:
     """Check which nodes are reachable and return their IPs."""
+
     reachable: dict[NodeId, set[str]] = {}
-    async with create_task_group() as tg:
+
+    # these are intentionally httpx's defaults so we can tune them later
+    timeout = httpx.Timeout(timeout=5.0)
+    limits = httpx.Limits(
+        max_connections=100,
+        max_keepalive_connections=20,
+        keepalive_expiry=5,
+    )
+
+    async with (
+        httpx.AsyncClient(timeout=timeout, limits=limits) as client,
+        create_task_group() as tg,
+    ):
         for node in topology.list_nodes():
             if not node.node_profile:
                 continue
+            if node.node_id == self_node_id:
+                continue
             for iface in node.node_profile.network_interfaces:
                 tg.start_soon(
                     check_reachability,
                     iface.ip_address,
                     node.node_id,
-                    self_node_id,
                     reachable,
+                    client,
                 )
 
     return reachable