Merge remote-tracking branch 'origin/main' into releases/v1.0.67

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

.github/actions/typecheck/action.yml

@@ -0,0 +1,12 @@
+name: Type Check
+
+description: "Run type checker"
+
+runs:
+  using: "composite"
+  steps:
+    - name: Run type checker
+      run: |
+        nix --extra-experimental-features nix-command --extra-experimental-features flakes develop -c just sync
+        nix --extra-experimental-features nix-command --extra-experimental-features flakes develop -c just check
+      shell: bash

.github/workflows/pipeline.yml

@@ -26,14 +26,73 @@ jobs:
           name: exo
           authToken: "${{ secrets.CACHIX_AUTH_TOKEN }}"
 
-      - name: Load nix develop environment
-        run: nix run github:nicknovitski/nix-develop/v1
+      - name: Configure git user
+        run: |
+          git config --local user.email "github-actions@users.noreply.github.com"
+          git config --local user.name  "github-actions bot"
+        shell: bash
 
-      - name: Sync dependencies
-        run: uv sync --all-packages
+      - name: Pull LFS files
+        run: |
+          echo "Pulling Git LFS files..."
+          git lfs pull
+        shell: bash
 
-      - name: Run type checker
-        run: uv run basedpyright --project pyproject.toml
+      - name: Setup Nix Environment
+        run: |
+          echo "Checking for nix installation..."
+          
+          # Check if nix binary exists directly
+          if [ -f /nix/var/nix/profiles/default/bin/nix ]; then
+            echo "Found nix binary at /nix/var/nix/profiles/default/bin/nix"
+            export PATH="/nix/var/nix/profiles/default/bin:$PATH"
+            echo "PATH=$PATH" >> $GITHUB_ENV
+            nix --version
+          elif [ -f /nix/var/nix/profiles/default/etc/profile.d/nix-daemon.sh ]; then
+            echo "Found nix profile script, sourcing..."
+            source /nix/var/nix/profiles/default/etc/profile.d/nix-daemon.sh
+            nix --version
+          elif command -v nix >/dev/null 2>&1; then
+            echo "Nix already in PATH"
+            nix --version
+          else
+            echo "Nix not found. Debugging info:"
+            echo "Contents of /nix/var/nix/profiles/default/:"
+            ls -la /nix/var/nix/profiles/default/ 2>/dev/null || echo "Directory not found"
+            echo "Contents of /nix/var/nix/profiles/default/bin/:"
+            ls -la /nix/var/nix/profiles/default/bin/ 2>/dev/null || echo "Directory not found"
+            exit 1
+          fi
+        shell: bash
+
+      - name: Configure basedpyright include for local MLX
+        run: |
+          RUNNER_LABELS='${{ toJSON(runner.labels) }}'
+          if echo "$RUNNER_LABELS" | grep -q "local_mlx"; then
+            if [ -d "/Users/Shared/mlx" ]; then
+              echo "Updating [tool.basedpyright].include to use /Users/Shared/mlx"
+              awk '
+                BEGIN { in=0 }
+                /^\[tool\.basedpyright\]/ { in=1; print; next }
+                in && /^\[/ { in=0 }  # next section
+                in && /^[ \t]*include[ \t]*=/ {
+                  print "include = [\"/Users/Shared/mlx\"]"
+                  next
+                }
+                { print }
+              ' pyproject.toml > pyproject.toml.tmp && mv pyproject.toml.tmp pyproject.toml
+
+              echo "New [tool.basedpyright] section:"
+              sed -n '/^\[tool\.basedpyright\]/,/^\[/p' pyproject.toml | sed '$d' || true
+            else
+              echo "local_mlx tag present but /Users/Shared/mlx not found; leaving pyproject unchanged."
+            fi
+          else
+            echo "Runner does not have 'local_mlx' tag; leaving pyproject unchanged."
+          fi
+        shell: bash
+
+      - uses: ./.github/actions/typecheck
 
   nix:
     name: Build and check (${{ matrix.system }})
@@ -64,63 +123,6 @@ jobs:
           name: exo
           authToken: "${{ secrets.CACHIX_AUTH_TOKEN }}"
 
-      - name: Build Metal packages (macOS only)
-        if: runner.os == 'macOS'
-        run: |
-          # Try to build metal-toolchain first (may succeed via cachix cache hit)
-          if nix build .#metal-toolchain 2>/dev/null; then
-            echo "metal-toolchain built successfully (likely cache hit)"
-          else
-            echo "metal-toolchain build failed, extracting from Xcode..."
-
-            NAR_HASH="sha256-ayR5mXN4sZAddwKEG2OszGRF93k9ZFc7H0yi2xbylQw="
-            NAR_NAME="metal-toolchain-17C48.nar"
-
-            # Use RUNNER_TEMP to avoid /tmp symlink issues on macOS
-            WORK_DIR="${RUNNER_TEMP}/metal-work"
-            mkdir -p "$WORK_DIR"
-
-            # Download the Metal toolchain component
-            xcodebuild -downloadComponent MetalToolchain
-
-            # Find and mount the DMG
-            DMG_PATH=$(find /System/Library/AssetsV2/com_apple_MobileAsset_MetalToolchain -name '*.dmg' 2>/dev/null | head -1)
-            if [ -z "$DMG_PATH" ]; then
-              echo "Error: Could not find Metal toolchain DMG"
-              exit 1
-            fi
-
-            echo "Found DMG at: $DMG_PATH"
-            hdiutil attach "$DMG_PATH" -mountpoint "${WORK_DIR}/metal-dmg"
-
-            # Copy the toolchain
-            cp -R "${WORK_DIR}/metal-dmg/Metal.xctoolchain" "${WORK_DIR}/metal-export"
-            hdiutil detach "${WORK_DIR}/metal-dmg"
-
-            # Create NAR and add to store
-            nix nar pack "${WORK_DIR}/metal-export" > "${WORK_DIR}/${NAR_NAME}"
-            STORE_PATH=$(nix store add --mode flat "${WORK_DIR}/${NAR_NAME}")
-            echo "Added NAR to store: $STORE_PATH"
-
-            # Verify the hash matches
-            ACTUAL_HASH=$(nix hash file "${WORK_DIR}/${NAR_NAME}")
-            if [ "$ACTUAL_HASH" != "$NAR_HASH" ]; then
-              echo "Warning: NAR hash mismatch!"
-              echo "Expected: $NAR_HASH"
-              echo "Actual:   $ACTUAL_HASH"
-              echo "The metal-toolchain.nix may need updating"
-            fi
-
-            # Clean up
-            rm -rf "$WORK_DIR"
-
-            # Retry the build now that NAR is in store
-            nix build .#metal-toolchain
-          fi
-
-          # Build mlx (depends on metal-toolchain)
-          nix build .#mlx
-
       - name: Build all Nix outputs
         run: |
           nix flake show --json | jq -r '
@@ -132,14 +134,3 @@ jobs:
 
       - name: Run nix flake check
         run: nix flake check
-
-      - name: Run pytest (macOS only)
-        if: runner.os == 'macOS'
-        run: |
-          # Build the test environment (requires relaxed sandbox for uv2nix on macOS)
-          TEST_ENV=$(nix build '.#exo-test-env' --option sandbox relaxed --print-out-paths)
-
-          # Run pytest outside sandbox (needs GPU access for MLX)
-          export HOME="$RUNNER_TEMP"
-          export EXO_TESTS=1
-          $TEST_ENV/bin/python -m pytest src -m "not slow" --import-mode=importlib

README.md

@@ -5,7 +5,7 @@
   <img alt="exo logo" src="/docs/imgs/exo-logo-transparent.png" width="50%" height="50%">
 </picture>
 
-exo: Run frontier AI locally. Maintained by [exo labs](https://x.com/exolabs).
+exo: Run your own AI cluster at home with everyday devices. Maintained by [exo labs](https://x.com/exolabs).
 
 <p align="center">
   <a href="https://discord.gg/TJ4P57arEm" target="_blank" rel="noopener noreferrer"><img src="https://img.shields.io/badge/Discord-Join%20Server-5865F2?logo=discord&logoColor=white" alt="Discord"></a>
@@ -107,10 +107,6 @@ uv run exo
 
 This starts the exo dashboard and API at http://localhost:52415/
 
-
-*Please view the section on RDMA to enable this feature on MacOS >=26.2!*
-
-
 ### Run from Source (Linux)
 
 **Prerequisites:**
@@ -234,7 +230,7 @@ This removes:
 
 RDMA is a new capability added to macOS 26.2. It works on any Mac with Thunderbolt 5 (M4 Pro Mac Mini, M4 Max Mac Studio, M4 Max MacBook Pro, M3 Ultra Mac Studio).
 
-Please refer to the caveats for immediate troubleshooting.
+Note that on Mac Studio, you cannot use the Thunderbolt 5 port next to the Ethernet port.
 
 To enable RDMA on macOS, follow these steps:
 
@@ -251,14 +247,6 @@ To enable RDMA on macOS, follow these steps:
 
 After that, RDMA will be enabled in macOS and exo will take care of the rest.
 
-**Important Caveats**
-
-1. Devices that wish to be part of an RDMA cluster must be connected to all other devices in the cluster.
-2. The cables must support TB5.
-3. On a Mac Studio, you cannot use the Thunderbolt 5 port next to the Ethernet port.
-4. If running from source, please use the script found at `tmp/set_rdma_network_config.sh`, which will disable Thunderbolt Bridge and set dhcp on each RDMA port.
-5. RDMA ports may be unable to discover each other on different versions of MacOS. Please ensure that OS versions match exactly (even beta version numbers) on all devices.
-
 ---
 
 ### Using the API

app/EXO/EXO.xcodeproj/project.pbxproj

@@ -342,8 +342,6 @@
 				SDKROOT = macosx;
 				SWIFT_ACTIVE_COMPILATION_CONDITIONS = "DEBUG $(inherited)";
 				SWIFT_OPTIMIZATION_LEVEL = "-Onone";
-				SWIFT_TREAT_WARNINGS_AS_ERRORS = YES;
-				GCC_TREAT_WARNINGS_AS_ERRORS = YES;
 			};
 			name = Debug;
 		};
@@ -399,8 +397,6 @@
 				MTL_FAST_MATH = YES;
 				SDKROOT = macosx;
 				SWIFT_COMPILATION_MODE = wholemodule;
-				SWIFT_TREAT_WARNINGS_AS_ERRORS = YES;
-				GCC_TREAT_WARNINGS_AS_ERRORS = YES;
 			};
 			name = Release;
 		};

app/EXO/EXO/EXOApp.swift

@@ -225,7 +225,7 @@ private final class ExoUpdaterDelegate: NSObject, SPUUpdaterDelegate {
         }
     }
 
-    nonisolated private func showNotification(title: String, body: String) {
+    private func showNotification(title: String, body: String) {
         let center = UNUserNotificationCenter.current()
         let content = UNMutableNotificationContent()
         content.title = title

app/EXO/EXO/Services/NetworkSetupHelper.swift

@@ -19,7 +19,7 @@ enum NetworkSetupHelper {
         set -euo pipefail
 
         # Wait for macOS to finish network setup after boot
-        sleep 20
+        sleep 30
 
         PREFS="/Library/Preferences/SystemConfiguration/preferences.plist"
 
@@ -244,11 +244,11 @@ enum NetworkSetupHelper {
         rm -f "$LOG_OUT" "$LOG_ERR"
 
         # Switch back to Automatic network location
-        networksetup -switchtolocation Automatic >/dev/null 2>&1 || true
+        networksetup -switchtolocation Automatic 2>/dev/null || true
 
         # Delete the exo network location if it exists
-        networksetup -listlocations 2>/dev/null | grep -q '^exo$' && {
-          networksetup -deletelocation exo >/dev/null 2>&1 || true
+        networksetup -listlocations | grep -q '^exo$' && {
+          networksetup -deletelocation exo 2>/dev/null || true
         } || true
 
         # Re-enable any Thunderbolt Bridge service if it exists
@@ -258,12 +258,12 @@ enum NetworkSetupHelper {
           tb_devices=$(networksetup -listallhardwareports 2>/dev/null | awk '
             /^Hardware Port:/ { port = tolower(substr($0, 16)) }
             /^Device:/ { if (port ~ /thunderbolt/) print substr($0, 9) }
-          ') || true
+          ')
           [ -z "$tb_devices" ] && return 0
 
           # For each bridge device, check if it contains Thunderbolt interfaces
           for bridge in bridge0 bridge1 bridge2; do
-            members=$(ifconfig "$bridge" 2>/dev/null | awk '/member:/ {print $2}') || true
+            members=$(ifconfig "$bridge" 2>/dev/null | awk '/member:/ {print $2}')
             [ -z "$members" ] && continue
 
             for tb_dev in $tb_devices; do
@@ -272,7 +272,7 @@ enum NetworkSetupHelper {
                 service_name=$(networksetup -listnetworkserviceorder 2>/dev/null | awk -v dev="$bridge" '
                   /^\\([0-9*]/ { gsub(/^\\([0-9*]+\\) /, ""); svc = $0 }
                   /Device:/ && $0 ~ dev { print svc; exit }
-                ') || true
+                ')
                 if [ -n "$service_name" ]; then
                   networksetup -setnetworkserviceenabled "$service_name" on 2>/dev/null || true
                   return 0
@@ -280,9 +280,8 @@ enum NetworkSetupHelper {
               fi
             done
           done
-          return 0
         }
-        find_and_enable_thunderbolt_bridge || true
+        find_and_enable_thunderbolt_bridge
 
         echo "EXO network components removed successfully"
         """

app/EXO/EXO/Services/ThunderboltBridgeService.swift

@@ -127,24 +127,21 @@ final class ThunderboltBridgeService: ObservableObject {
 
         // 2. Request specific network configuration rights
         let rightName = "system.services.systemconfiguration.network"
-        status = rightName.withCString { nameCString in
-            var item = AuthorizationItem(
-                name: nameCString,
-                valueLength: 0,
-                value: nil,
-                flags: 0
-            )
-            return withUnsafeMutablePointer(to: &item) { itemPointer in
-                var rights = AuthorizationRights(count: 1, items: itemPointer)
-                return AuthorizationCopyRights(
-                    authRef,
-                    &rights,
-                    nil,
-                    [.extendRights, .interactionAllowed],
-                    nil
-                )
-            }
-        }
+        var item = AuthorizationItem(
+            name: rightName,
+            valueLength: 0,
+            value: nil,
+            flags: 0
+        )
+        var rights = AuthorizationRights(count: 1, items: &item)
+
+        status = AuthorizationCopyRights(
+            authRef,
+            &rights,
+            nil,
+            [.extendRights, .interactionAllowed],
+            nil
+        )
         guard status == errAuthorizationSuccess else {
             if status == errAuthorizationCanceled {
                 throw ThunderboltBridgeError.authorizationCanceled

app/EXO/uninstall-exo.sh

@@ -29,21 +29,21 @@ YELLOW='\033[1;33m'
 NC='\033[0m' # No Color
 
 echo_info() {
-  echo -e "${GREEN}[INFO]${NC} $1"
+    echo -e "${GREEN}[INFO]${NC} $1"
 }
 
 echo_warn() {
-  echo -e "${YELLOW}[WARN]${NC} $1"
+    echo -e "${YELLOW}[WARN]${NC} $1"
 }
 
 echo_error() {
-  echo -e "${RED}[ERROR]${NC} $1"
+    echo -e "${RED}[ERROR]${NC} $1"
 }
 
 # Check if running as root
 if [[ $EUID -ne 0 ]]; then
-  echo_error "This script must be run as root (use sudo)"
-  exit 1
+    echo_error "This script must be run as root (use sudo)"
+    exit 1
 fi
 
 echo ""
@@ -55,64 +55,64 @@ echo ""
 # Unload the LaunchDaemon if running
 echo_info "Stopping network setup daemon..."
 if launchctl list | grep -q "$LABEL"; then
-  launchctl bootout system/"$LABEL" 2>/dev/null || true
-  echo_info "Daemon stopped"
+    launchctl bootout system/"$LABEL" 2>/dev/null || true
+    echo_info "Daemon stopped"
 else
-  echo_warn "Daemon was not running"
+    echo_warn "Daemon was not running"
 fi
 
 # Remove LaunchDaemon plist
-if [[ -f $PLIST_DEST ]]; then
-  rm -f "$PLIST_DEST"
-  echo_info "Removed LaunchDaemon plist"
+if [[ -f "$PLIST_DEST" ]]; then
+    rm -f "$PLIST_DEST"
+    echo_info "Removed LaunchDaemon plist"
 else
-  echo_warn "LaunchDaemon plist not found (already removed?)"
+    echo_warn "LaunchDaemon plist not found (already removed?)"
 fi
 
 # Remove the script and parent directory
-if [[ -f $SCRIPT_DEST ]]; then
-  rm -f "$SCRIPT_DEST"
-  echo_info "Removed network setup script"
+if [[ -f "$SCRIPT_DEST" ]]; then
+    rm -f "$SCRIPT_DEST"
+    echo_info "Removed network setup script"
 else
-  echo_warn "Network setup script not found (already removed?)"
+    echo_warn "Network setup script not found (already removed?)"
 fi
 
 # Remove EXO directory if empty
 if [[ -d "/Library/Application Support/EXO" ]]; then
-  rmdir "/Library/Application Support/EXO" 2>/dev/null &&
-    echo_info "Removed EXO support directory" ||
-    echo_warn "EXO support directory not empty, leaving in place"
+    rmdir "/Library/Application Support/EXO" 2>/dev/null && \
+        echo_info "Removed EXO support directory" || \
+        echo_warn "EXO support directory not empty, leaving in place"
 fi
 
 # Remove log files
-if [[ -f $LOG_OUT ]] || [[ -f $LOG_ERR ]]; then
-  rm -f "$LOG_OUT" "$LOG_ERR"
-  echo_info "Removed log files"
+if [[ -f "$LOG_OUT" ]] || [[ -f "$LOG_ERR" ]]; then
+    rm -f "$LOG_OUT" "$LOG_ERR"
+    echo_info "Removed log files"
 else
-  echo_warn "Log files not found (already removed?)"
+    echo_warn "Log files not found (already removed?)"
 fi
 
 # Switch back to Automatic network location
 echo_info "Restoring network configuration..."
 if networksetup -listlocations | grep -q "^Automatic$"; then
-  networksetup -switchtolocation Automatic 2>/dev/null || true
-  echo_info "Switched to Automatic network location"
+    networksetup -switchtolocation Automatic 2>/dev/null || true
+    echo_info "Switched to Automatic network location"
 else
-  echo_warn "Automatic network location not found"
+    echo_warn "Automatic network location not found"
 fi
 
 # Delete the exo network location if it exists
 if networksetup -listlocations | grep -q "^exo$"; then
-  networksetup -deletelocation exo 2>/dev/null || true
-  echo_info "Deleted 'exo' network location"
+    networksetup -deletelocation exo 2>/dev/null || true
+    echo_info "Deleted 'exo' network location"
 else
-  echo_warn "'exo' network location not found (already removed?)"
+    echo_warn "'exo' network location not found (already removed?)"
 fi
 
 # Re-enable Thunderbolt Bridge if it exists
 if networksetup -listnetworkservices 2>/dev/null | grep -q "Thunderbolt Bridge"; then
-  networksetup -setnetworkserviceenabled "Thunderbolt Bridge" on 2>/dev/null || true
-  echo_info "Re-enabled Thunderbolt Bridge"
+    networksetup -setnetworkserviceenabled "Thunderbolt Bridge" on 2>/dev/null || true
+    echo_info "Re-enabled Thunderbolt Bridge"
 fi
 
 # Note about launch at login registration
@@ -124,14 +124,14 @@ echo_warn "  System Settings → General → Login Items → Remove EXO"
 # Check if EXO.app exists in common locations
 APP_FOUND=false
 for app_path in "/Applications/EXO.app" "$HOME/Applications/EXO.app"; do
-  if [[ -d $app_path ]]; then
-    if [[ $APP_FOUND == false ]]; then
-      echo ""
-      APP_FOUND=true
+    if [[ -d "$app_path" ]]; then
+        if [[ "$APP_FOUND" == false ]]; then
+            echo ""
+            APP_FOUND=true
+        fi
+        echo_warn "EXO.app found at: $app_path"
+        echo_warn "You may want to move it to Trash manually."
     fi
-    echo_warn "EXO.app found at: $app_path"
-    echo_warn "You may want to move it to Trash manually."
-  fi
 done
 
 echo ""
@@ -151,3 +151,4 @@ echo ""
 echo "Manual step required:"
 echo "  Remove EXO from Login Items in System Settings → General → Login Items"
 echo ""
+

dashboard/package-lock.json

@@ -865,6 +865,7 @@
 			"integrity": "sha512-oH8tXw7EZnie8FdOWYrF7Yn4IKrqTFHhXvl8YxXxbKwTMcD/5NNCryUSEXRk2ZR4ojnub0P8rNrsVGHXWqIDtA==",
 			"dev": true,
 			"license": "MIT",
+			"peer": true,
 			"dependencies": {
 				"@standard-schema/spec": "^1.0.0",
 				"@sveltejs/acorn-typescript": "^1.0.5",
@@ -904,6 +905,7 @@
 			"integrity": "sha512-Y1Cs7hhTc+a5E9Va/xwKlAJoariQyHY+5zBgCZg4PFWNYQ1nMN9sjK1zhw1gK69DuqVP++sht/1GZg1aRwmAXQ==",
 			"dev": true,
 			"license": "MIT",
+			"peer": true,
 			"dependencies": {
 				"@sveltejs/vite-plugin-svelte-inspector": "^4.0.1",
 				"debug": "^4.4.1",
@@ -1520,6 +1522,7 @@
 			"integrity": "sha512-LCCV0HdSZZZb34qifBsyWlUmok6W7ouER+oQIGBScS8EsZsQbrtFTUrDX4hOl+CS6p7cnNC4td+qrSVGSCTUfQ==",
 			"dev": true,
 			"license": "MIT",
+			"peer": true,
 			"dependencies": {
 				"undici-types": "~6.21.0"
 			}
@@ -1529,6 +1532,7 @@
 			"resolved": "https://registry.npmjs.org/acorn/-/acorn-8.15.0.tgz",
 			"integrity": "sha512-NZyJarBfL7nWwIq+FDL6Zp/yHEhePMNnnJ0y3qfieCrmNvYct8uvtiV41UvlSe6apAfk0fY1FbWx+NwfmpvtTg==",
 			"license": "MIT",
+			"peer": true,
 			"bin": {
 				"acorn": "bin/acorn"
 			},
@@ -1941,6 +1945,7 @@
 			"integrity": "sha512-fmTRWbNMmsmWq6xJV8D19U/gw/bwrHfNXxrIN+HfZgnzqTHp9jOmKMhsTUjXOJnZOdZY9Q28y4yebKzqDKlxlQ==",
 			"dev": true,
 			"license": "ISC",
+			"peer": true,
 			"engines": {
 				"node": ">=12"
 			}
@@ -2648,6 +2653,7 @@
 			"integrity": "sha512-5gTmgEY/sqK6gFXLIsQNH19lWb4ebPDLA4SdLP7dsWkIXHWlG66oPuVvXSGFPppYZz8ZDZq0dYYrbHfBCVUb1Q==",
 			"dev": true,
 			"license": "MIT",
+			"peer": true,
 			"engines": {
 				"node": ">=12"
 			},
@@ -2690,6 +2696,7 @@
 			"integrity": "sha512-UOnG6LftzbdaHZcKoPFtOcCKztrQ57WkHDeRD9t/PTQtmT0NHSeWWepj6pS0z/N7+08BHFDQVUrfmfMRcZwbMg==",
 			"dev": true,
 			"license": "MIT",
+			"peer": true,
 			"bin": {
 				"prettier": "bin/prettier.cjs"
 			},
@@ -2862,6 +2869,7 @@
 			"resolved": "https://registry.npmjs.org/svelte/-/svelte-5.45.3.tgz",
 			"integrity": "sha512-ngKXNhNvwPzF43QqEhDOue7TQTrG09em1sd4HBxVF0Wr2gopAmdEWan+rgbdgK4fhBtSOTJO8bYU4chUG7VXZQ==",
 			"license": "MIT",
+			"peer": true,
 			"dependencies": {
 				"@jridgewell/remapping": "^2.3.4",
 				"@jridgewell/sourcemap-codec": "^1.5.0",
@@ -3006,6 +3014,7 @@
 			"integrity": "sha512-jl1vZzPDinLr9eUt3J/t7V6FgNEw9QjvBPdysz9KfQDD41fQrC2Y4vKQdiaUpFT4bXlb1RHhLpp8wtm6M5TgSw==",
 			"dev": true,
 			"license": "Apache-2.0",
+			"peer": true,
 			"bin": {
 				"tsc": "bin/tsc",
 				"tsserver": "bin/tsserver"
@@ -3027,6 +3036,7 @@
 			"integrity": "sha512-+Oxm7q9hDoLMyJOYfUYBuHQo+dkAloi33apOPP56pzj+vsdJDzr+j1NISE5pyaAuKL4A3UD34qd0lx5+kfKp2g==",
 			"dev": true,
 			"license": "MIT",
+			"peer": true,
 			"dependencies": {
 				"esbuild": "^0.25.0",
 				"fdir": "^6.4.4",

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

@@ -173,11 +173,6 @@ export interface PlacementPreviewResponse {
   previews: PlacementPreview[];
 }
 
-interface ImageApiResponse {
-  created: number;
-  data: Array<{ b64_json?: string; url?: string }>;
-}
-
 interface RawStateResponse {
   topology?: RawTopology;
   instances?: Record<
@@ -2100,137 +2095,107 @@ class AppStore {
         throw new Error(`API error: ${response.status} - ${errorText}`);
       }
 
-      // Streaming requires both stream=true AND partialImages > 0
-      const isStreaming = params.stream && params.partialImages > 0;
-
-      if (!isStreaming) {
-        // Non-streaming: parse JSON response directly
-        const jsonResponse = (await response.json()) as ImageApiResponse;
-        const format = params.outputFormat || "png";
-        const mimeType = `image/${format}`;
-
-        const attachments: MessageAttachment[] = jsonResponse.data
-          .filter((img) => img.b64_json)
-          .map((img, index) => ({
-            type: "generated-image" as const,
-            name: `generated-image-${index + 1}.${format}`,
-            preview: `data:${mimeType};base64,${img.b64_json}`,
-            mimeType,
-          }));
-
-        this.updateConversationMessage(
-          targetConversationId,
-          assistantMessage.id,
-          (msg) => {
-            msg.content = "";
-            msg.attachments = attachments;
-          },
-        );
-        this.syncActiveMessagesIfNeeded(targetConversationId);
-      } else {
-        // Streaming mode: use SSE parser
-        const reader = response.body?.getReader();
-        if (!reader) {
-          throw new Error("No response body");
-        }
-
-        interface ImageGenerationChunk {
-          data?: { b64_json?: string };
-          format?: string;
-          type?: "partial" | "final";
-          image_index?: number;
-          partial_index?: number;
-          total_partials?: number;
-        }
-
-        const numImages = params.numImages;
-
-        await this.parseSSEStream<ImageGenerationChunk>(
-          reader,
-          targetConversationId,
-          (parsed) => {
-            const imageData = parsed.data?.b64_json;
-
-            if (imageData) {
-              const format = parsed.format || "png";
-              const mimeType = `image/${format}`;
-              const imageIndex = parsed.image_index ?? 0;
-
-              if (parsed.type === "partial") {
-                // Update with partial image and progress
-                const partialNum = (parsed.partial_index ?? 0) + 1;
-                const totalPartials = parsed.total_partials ?? 3;
-                const progressText =
-                  numImages > 1
-                    ? `Generating image ${imageIndex + 1}/${numImages}... ${partialNum}/${totalPartials}`
-                    : `Generating... ${partialNum}/${totalPartials}`;
-
-                const partialAttachment: MessageAttachment = {
-                  type: "generated-image",
-                  name: `generated-image.${format}`,
-                  preview: `data:${mimeType};base64,${imageData}`,
-                  mimeType,
-                };
-
-                this.updateConversationMessage(
-                  targetConversationId,
-                  assistantMessage.id,
-                  (msg) => {
-                    msg.content = progressText;
-                    if (imageIndex === 0) {
-                      // First image - safe to replace attachments with partial preview
-                      msg.attachments = [partialAttachment];
-                    } else {
-                      // Subsequent images - keep existing finals, show partial at current position
-                      const existingAttachments = msg.attachments || [];
-                      // Keep only the completed final images (up to current imageIndex)
-                      const finals = existingAttachments.slice(0, imageIndex);
-                      msg.attachments = [...finals, partialAttachment];
-                    }
-                  },
-                );
-              } else if (parsed.type === "final") {
-                // Final image - replace partial at this position
-                const newAttachment: MessageAttachment = {
-                  type: "generated-image",
-                  name: `generated-image-${imageIndex + 1}.${format}`,
-                  preview: `data:${mimeType};base64,${imageData}`,
-                  mimeType,
-                };
-
-                this.updateConversationMessage(
-                  targetConversationId,
-                  assistantMessage.id,
-                  (msg) => {
-                    if (imageIndex === 0) {
-                      // First final image - replace any partial preview
-                      msg.attachments = [newAttachment];
-                    } else {
-                      // Subsequent images - keep previous finals, replace partial at current position
-                      const existingAttachments = msg.attachments || [];
-                      // Slice keeps indices 0 to imageIndex-1 (the previous final images)
-                      const previousFinals = existingAttachments.slice(
-                        0,
-                        imageIndex,
-                      );
-                      msg.attachments = [...previousFinals, newAttachment];
-                    }
-
-                    // Update progress message for multiple images
-                    if (numImages > 1 && imageIndex < numImages - 1) {
-                      msg.content = `Generating image ${imageIndex + 2}/${numImages}...`;
-                    } else {
-                      msg.content = "";
-                    }
-                  },
-                );
-              }
-
-              this.syncActiveMessagesIfNeeded(targetConversationId);
-            }
-          },
-        );
+      const reader = response.body?.getReader();
+      if (!reader) {
+        throw new Error("No response body");
       }
+
+      interface ImageGenerationChunk {
+        data?: { b64_json?: string };
+        format?: string;
+        type?: "partial" | "final";
+        image_index?: number;
+        partial_index?: number;
+        total_partials?: number;
+      }
+
+      const numImages = params.numImages;
+
+      await this.parseSSEStream<ImageGenerationChunk>(
+        reader,
+        targetConversationId,
+        (parsed) => {
+          const imageData = parsed.data?.b64_json;
+
+          if (imageData) {
+            const format = parsed.format || "png";
+            const mimeType = `image/${format}`;
+            const imageIndex = parsed.image_index ?? 0;
+
+            if (parsed.type === "partial") {
+              // Update with partial image and progress
+              const partialNum = (parsed.partial_index ?? 0) + 1;
+              const totalPartials = parsed.total_partials ?? 3;
+              const progressText =
+                numImages > 1
+                  ? `Generating image ${imageIndex + 1}/${numImages}... ${partialNum}/${totalPartials}`
+                  : `Generating... ${partialNum}/${totalPartials}`;
+
+              const partialAttachment: MessageAttachment = {
+                type: "generated-image",
+                name: `generated-image.${format}`,
+                preview: `data:${mimeType};base64,${imageData}`,
+                mimeType,
+              };
+
+              this.updateConversationMessage(
+                targetConversationId,
+                assistantMessage.id,
+                (msg) => {
+                  msg.content = progressText;
+                  if (imageIndex === 0) {
+                    // First image - safe to replace attachments with partial preview
+                    msg.attachments = [partialAttachment];
+                  } else {
+                    // Subsequent images - keep existing finals, show partial at current position
+                    const existingAttachments = msg.attachments || [];
+                    // Keep only the completed final images (up to current imageIndex)
+                    const finals = existingAttachments.slice(0, imageIndex);
+                    msg.attachments = [...finals, partialAttachment];
+                  }
+                },
+              );
+            } else if (parsed.type === "final") {
+              // Final image - replace partial at this position
+              const newAttachment: MessageAttachment = {
+                type: "generated-image",
+                name: `generated-image-${imageIndex + 1}.${format}`,
+                preview: `data:${mimeType};base64,${imageData}`,
+                mimeType,
+              };
+
+              this.updateConversationMessage(
+                targetConversationId,
+                assistantMessage.id,
+                (msg) => {
+                  if (imageIndex === 0) {
+                    // First final image - replace any partial preview
+                    msg.attachments = [newAttachment];
+                  } else {
+                    // Subsequent images - keep previous finals, replace partial at current position
+                    const existingAttachments = msg.attachments || [];
+                    // Slice keeps indices 0 to imageIndex-1 (the previous final images)
+                    const previousFinals = existingAttachments.slice(
+                      0,
+                      imageIndex,
+                    );
+                    msg.attachments = [...previousFinals, newAttachment];
+                  }
+
+                  // Update progress message for multiple images
+                  if (numImages > 1 && imageIndex < numImages - 1) {
+                    msg.content = `Generating image ${imageIndex + 2}/${numImages}...`;
+                  } else {
+                    msg.content = "";
+                  }
+                },
+              );
+            }
+
+            this.syncActiveMessagesIfNeeded(targetConversationId);
+          }
+        },
+      );
     } catch (error) {
       console.error("Error generating image:", error);
       this.handleStreamingError(
@@ -2378,98 +2343,69 @@ class AppStore {
         throw new Error(`API error: ${apiResponse.status} - ${errorText}`);
       }
 
-      // Streaming requires both stream=true AND partialImages > 0
-      const isStreaming = params.stream && params.partialImages > 0;
-
-      if (!isStreaming) {
-        // Non-streaming: parse JSON response directly
-        const jsonResponse = (await apiResponse.json()) as ImageApiResponse;
-        const format = params.outputFormat || "png";
-        const mimeType = `image/${format}`;
-        const attachments: MessageAttachment[] = jsonResponse.data
-          .filter((img) => img.b64_json)
-          .map((img) => ({
-            type: "generated-image" as const,
-            name: `edited-image.${format}`,
-            preview: `data:${mimeType};base64,${img.b64_json}`,
-            mimeType,
-          }));
-
-        this.updateConversationMessage(
-          targetConversationId,
-          assistantMessage.id,
-          (msg) => {
-            msg.content = "";
-            msg.attachments = attachments;
-          },
-        );
-        this.syncActiveMessagesIfNeeded(targetConversationId);
-      } else {
-        // Streaming mode: use SSE parser
-        const reader = apiResponse.body?.getReader();
-        if (!reader) {
-          throw new Error("No response body");
-        }
-
-        interface ImageEditChunk {
-          data?: { b64_json?: string };
-          format?: string;
-          type?: "partial" | "final";
-          partial_index?: number;
-          total_partials?: number;
-        }
-
-        await this.parseSSEStream<ImageEditChunk>(
-          reader,
-          targetConversationId,
-          (parsed) => {
-            const imageData = parsed.data?.b64_json;
-
-            if (imageData) {
-              const format = parsed.format || "png";
-              const mimeType = `image/${format}`;
-              if (parsed.type === "partial") {
-                // Update with partial image and progress
-                const partialNum = (parsed.partial_index ?? 0) + 1;
-                const totalPartials = parsed.total_partials ?? 3;
-                this.updateConversationMessage(
-                  targetConversationId,
-                  assistantMessage.id,
-                  (msg) => {
-                    msg.content = `Editing... ${partialNum}/${totalPartials}`;
-                    msg.attachments = [
-                      {
-                        type: "generated-image",
-                        name: `edited-image.${format}`,
-                        preview: `data:${mimeType};base64,${imageData}`,
-                        mimeType,
-                      },
-                    ];
-                  },
-                );
-              } else if (parsed.type === "final") {
-                // Final image
-                this.updateConversationMessage(
-                  targetConversationId,
-                  assistantMessage.id,
-                  (msg) => {
-                    msg.content = "";
-                    msg.attachments = [
-                      {
-                        type: "generated-image",
-                        name: `edited-image.${format}`,
-                        preview: `data:${mimeType};base64,${imageData}`,
-                        mimeType,
-                      },
-                    ];
-                  },
-                );
-              }
-              this.syncActiveMessagesIfNeeded(targetConversationId);
-            }
-          },
-        );
+      const reader = apiResponse.body?.getReader();
+      if (!reader) {
+        throw new Error("No response body");
       }
+
+      interface ImageEditChunk {
+        data?: { b64_json?: string };
+        format?: string;
+        type?: "partial" | "final";
+        partial_index?: number;
+        total_partials?: number;
+      }
+
+      await this.parseSSEStream<ImageEditChunk>(
+        reader,
+        targetConversationId,
+        (parsed) => {
+          const imageData = parsed.data?.b64_json;
+
+          if (imageData) {
+            const format = parsed.format || "png";
+            const mimeType = `image/${format}`;
+            if (parsed.type === "partial") {
+              // Update with partial image and progress
+              const partialNum = (parsed.partial_index ?? 0) + 1;
+              const totalPartials = parsed.total_partials ?? 3;
+              this.updateConversationMessage(
+                targetConversationId,
+                assistantMessage.id,
+                (msg) => {
+                  msg.content = `Editing... ${partialNum}/${totalPartials}`;
+                  msg.attachments = [
+                    {
+                      type: "generated-image",
+                      name: `edited-image.${format}`,
+                      preview: `data:${mimeType};base64,${imageData}`,
+                      mimeType,
+                    },
+                  ];
+                },
+              );
+            } else if (parsed.type === "final") {
+              // Final image
+              this.updateConversationMessage(
+                targetConversationId,
+                assistantMessage.id,
+                (msg) => {
+                  msg.content = "";
+                  msg.attachments = [
+                    {
+                      type: "generated-image",
+                      name: `edited-image.${format}`,
+                      preview: `data:${mimeType};base64,${imageData}`,
+                      mimeType,
+                    },
+                  ];
+                },
+              );
+            }
+            this.syncActiveMessagesIfNeeded(targetConversationId);
+          }
+        },
+      );
     } catch (error) {
       console.error("Error editing image:", error);
       this.handleStreamingError(

flake.lock

@@ -21,9 +21,7 @@
           "nixpkgs"
         ],
         "purescript-overlay": "purescript-overlay",
-        "pyproject-nix": [
-          "pyproject-nix"
-        ]
+        "pyproject-nix": "pyproject-nix"
       },
       "locked": {
         "lastModified": 1765953015,
@@ -151,44 +149,19 @@
         "type": "github"
       }
     },
-    "pyproject-build-systems": {
-      "inputs": {
-        "nixpkgs": [
-          "nixpkgs"
-        ],
-        "pyproject-nix": [
-          "pyproject-nix"
-        ],
-        "uv2nix": [
-          "uv2nix"
-        ]
-      },
-      "locked": {
-        "lastModified": 1763662255,
-        "narHash": "sha256-4bocaOyLa3AfiS8KrWjZQYu+IAta05u3gYZzZ6zXbT0=",
-        "owner": "pyproject-nix",
-        "repo": "build-system-pkgs",
-        "rev": "042904167604c681a090c07eb6967b4dd4dae88c",
-        "type": "github"
-      },
-      "original": {
-        "owner": "pyproject-nix",
-        "repo": "build-system-pkgs",
-        "type": "github"
-      }
-    },
     "pyproject-nix": {
       "inputs": {
         "nixpkgs": [
+          "dream2nix",
           "nixpkgs"
         ]
       },
       "locked": {
-        "lastModified": 1764134915,
-        "narHash": "sha256-xaKvtPx6YAnA3HQVp5LwyYG1MaN4LLehpQI8xEdBvBY=",
+        "lastModified": 1763017646,
+        "narHash": "sha256-Z+R2lveIp6Skn1VPH3taQIuMhABg1IizJd8oVdmdHsQ=",
         "owner": "pyproject-nix",
         "repo": "pyproject.nix",
-        "rev": "2c8df1383b32e5443c921f61224b198a2282a657",
+        "rev": "47bd6f296502842643078d66128f7b5e5370790c",
         "type": "github"
       },
       "original": {
@@ -205,10 +178,7 @@
         "flake-parts": "flake-parts",
         "nixpkgs": "nixpkgs",
         "nixpkgs-swift": "nixpkgs-swift",
-        "pyproject-build-systems": "pyproject-build-systems",
-        "pyproject-nix": "pyproject-nix",
-        "treefmt-nix": "treefmt-nix",
-        "uv2nix": "uv2nix"
+        "treefmt-nix": "treefmt-nix"
       }
     },
     "rust-analyzer-src": {
@@ -269,29 +239,6 @@
         "repo": "treefmt-nix",
         "type": "github"
       }
-    },
-    "uv2nix": {
-      "inputs": {
-        "nixpkgs": [
-          "nixpkgs"
-        ],
-        "pyproject-nix": [
-          "pyproject-nix"
-        ]
-      },
-      "locked": {
-        "lastModified": 1767701098,
-        "narHash": "sha256-CJhKZnWb3gumR9oTRjFvCg/6lYTGbZRU7xtvcyWIRwU=",
-        "owner": "pyproject-nix",
-        "repo": "uv2nix",
-        "rev": "9d357f0d2ce6f5f35ec7959d7e704452352eb4da",
-        "type": "github"
-      },
-      "original": {
-        "owner": "pyproject-nix",
-        "repo": "uv2nix",
-        "type": "github"
-      }
     }
   },
   "root": "root",

flake.nix

@@ -24,26 +24,6 @@
     dream2nix = {
       url = "github:nix-community/dream2nix";
       inputs.nixpkgs.follows = "nixpkgs";
-      inputs.pyproject-nix.follows = "pyproject-nix";
-    };
-
-    # Python packaging with uv2nix
-    pyproject-nix = {
-      url = "github:pyproject-nix/pyproject.nix";
-      inputs.nixpkgs.follows = "nixpkgs";
-    };
-
-    uv2nix = {
-      url = "github:pyproject-nix/uv2nix";
-      inputs.pyproject-nix.follows = "pyproject-nix";
-      inputs.nixpkgs.follows = "nixpkgs";
-    };
-
-    pyproject-build-systems = {
-      url = "github:pyproject-nix/build-system-pkgs";
-      inputs.pyproject-nix.follows = "pyproject-nix";
-      inputs.uv2nix.follows = "uv2nix";
-      inputs.nixpkgs.follows = "nixpkgs";
     };
 
     # Pinned nixpkgs for swift-format (swift is broken on x86_64-linux in newer nixpkgs)
@@ -68,7 +48,6 @@
         inputs.treefmt-nix.flakeModule
         ./dashboard/parts.nix
         ./rust/parts.nix
-        ./python/parts.nix
       ];
 
       perSystem =
@@ -79,11 +58,6 @@
           pkgsSwift = import inputs.nixpkgs-swift { inherit system; };
         in
         {
-          # Allow unfree for metal-toolchain (needed for Darwin Metal packages)
-          _module.args.pkgs = import inputs.nixpkgs {
-            inherit system;
-            config.allowUnfreePredicate = pkg: (pkg.pname or "") == "metal-toolchain";
-          };
           treefmt = {
             projectRootFile = "flake.nix";
             programs = {
@@ -105,24 +79,14 @@
                 enable = true;
                 package = pkgsSwift.swiftPackages.swift-format;
               };
-              shfmt.enable = true;
             };
           };
 
-          packages = lib.optionalAttrs pkgs.stdenv.hostPlatform.isDarwin (
-            let
-              uvLock = builtins.fromTOML (builtins.readFile ./uv.lock);
-              mlxPackage = builtins.head (builtins.filter (p: p.name == "mlx") uvLock.package);
-              uvLockMlxVersion = mlxPackage.version;
-            in
-            {
-              metal-toolchain = pkgs.callPackage ./nix/metal-toolchain.nix { };
-              mlx = pkgs.callPackage ./nix/mlx.nix {
-                metal-toolchain = self'.packages.metal-toolchain;
-                inherit uvLockMlxVersion;
-              };
-            }
-          );
+          checks.lint = pkgs.runCommand "lint-check" { } ''
+            export RUFF_CACHE_DIR="$TMPDIR/ruff-cache"
+            ${pkgs.ruff}/bin/ruff check ${inputs.self}/
+            touch $out
+          '';
 
           devShells.default = with pkgs; pkgs.mkShell {
             inputsFrom = [ self'.checks.cargo-build ];

nix/darwin-build-fixes.patch

@@ -1,79 +0,0 @@
-diff --git a/CMakeLists.txt b/CMakeLists.txt
-index 0ed30932..d8528132 100644
---- a/CMakeLists.txt
-+++ b/CMakeLists.txt
-@@ -177,11 +177,7 @@ if(MLX_BUILD_METAL)
-     add_compile_definitions(MLX_METAL_DEBUG)
-   endif()
-
--  # Throw an error if xcrun not found
--  execute_process(
--    COMMAND zsh "-c" "/usr/bin/xcrun -sdk macosx --show-sdk-version"
--    OUTPUT_VARIABLE MACOS_SDK_VERSION
--    OUTPUT_STRIP_TRAILING_WHITESPACE COMMAND_ERROR_IS_FATAL ANY)
-+  set(MACOS_SDK_VERSION @sdkVersion@)
-
-   if(${MACOS_SDK_VERSION} LESS 14.0)
-     message(
-@@ -199,11 +195,8 @@ if(MLX_BUILD_METAL)
-     endif()
-     set(XCRUN_FLAGS "-mmacosx-version-min=${CMAKE_OSX_DEPLOYMENT_TARGET}")
-   endif()
--  execute_process(
--    COMMAND
--      zsh "-c"
--      "echo \"__METAL_VERSION__\" | xcrun -sdk macosx metal ${XCRUN_FLAGS} -E -x metal -P - | tail -1 | tr -d '\n'"
--    OUTPUT_VARIABLE MLX_METAL_VERSION COMMAND_ERROR_IS_FATAL ANY)
-+  set(
-+    MLX_METAL_VERSION @metalVersion@)
-   FetchContent_Declare(metal_cpp URL ${METAL_CPP_URL})
-   FetchContent_MakeAvailable(metal_cpp)
-   target_include_directories(
-diff --git a/cmake/extension.cmake b/cmake/extension.cmake
-index 13db804a..5b385132 100644
---- a/cmake/extension.cmake
-+++ b/cmake/extension.cmake
-@@ -36,7 +36,7 @@ macro(mlx_build_metallib)
-   add_custom_command(
-     OUTPUT ${MTLLIB_BUILD_TARGET}
-     COMMAND
--      xcrun -sdk macosx metal
-+      metal -fmodules-cache-path=${CMAKE_BINARY_DIR}/metal-cache
-       "$<LIST:TRANSFORM,${MTLLIB_INCLUDE_DIRS},PREPEND,-I>"
-       ${MTLLIB_COMPILE_OPTIONS} ${MTLLIB_SOURCES} -o ${MTLLIB_BUILD_TARGET}
-     DEPENDS ${MTLLIB_DEPS} ${MTLLIB_SOURCES}
-diff --git a/mlx/backend/metal/kernels/CMakeLists.txt b/mlx/backend/metal/kernels/CMakeLists.txt
-index 262b0495..5c7446ad 100644
---- a/mlx/backend/metal/kernels/CMakeLists.txt
-+++ b/mlx/backend/metal/kernels/CMakeLists.txt
-@@ -29,7 +29,7 @@ function(build_kernel_base TARGET SRCFILE DEPS)
-                     "-mmacosx-version-min=${CMAKE_OSX_DEPLOYMENT_TARGET}")
-   endif()
-   add_custom_command(
--    COMMAND xcrun -sdk macosx metal ${METAL_FLAGS} -c ${SRCFILE}
-+    COMMAND metal -fmodules-cache-path=${CMAKE_BINARY_DIR}/metal-cache ${METAL_FLAGS} -c ${SRCFILE}
-             -I${PROJECT_SOURCE_DIR} -o ${TARGET}.air
-     DEPENDS ${SRCFILE} ${DEPS} ${BASE_HEADERS}
-     OUTPUT ${TARGET}.air
-@@ -170,7 +170,7 @@ endif()
-
- add_custom_command(
-   OUTPUT ${MLX_METAL_PATH}/mlx.metallib
--  COMMAND xcrun -sdk macosx metallib ${KERNEL_AIR} -o
-+  COMMAND metallib ${KERNEL_AIR} -o
-           ${MLX_METAL_PATH}/mlx.metallib
-   DEPENDS ${KERNEL_AIR}
-   COMMENT "Building mlx.metallib"
-diff --git a/mlx/backend/metal/make_compiled_preamble.sh b/mlx/backend/metal/make_compiled_preamble.sh
-index bb55ed3a..94ea7dd7 100644
---- a/mlx/backend/metal/make_compiled_preamble.sh
-+++ b/mlx/backend/metal/make_compiled_preamble.sh
-@@ -31,7 +31,7 @@ OUTPUT_FILE=${OUTPUT_DIR}/${SRC_NAME}.cpp
- mkdir -p "$OUTPUT_DIR"
-
- # Use the metal compiler to get a list of headers (with depth)
--CCC="xcrun -sdk macosx metal -x metal"
-+CCC="metal -x metal -fmodules-cache-path=${OUTPUT_DIR}/metal-cache"
- HDRS=$( $CCC -I"$SRC_DIR" -I"$JIT_INCLUDES" -DMLX_METAL_JIT -E -P -CC -C -H "$INPUT_FILE" $CFLAGS -w 2>&1 1>/dev/null )
-
- # Remove any included system frameworks (for MetalPerformancePrimitive headers)

nix/metal-toolchain.nix

@@ -1,56 +0,0 @@
-{ lib, stdenvNoCC, requireFile, nix }:
-
-let
-  narFile = requireFile {
-    name = "metal-toolchain-17C48.nar";
-    message = ''
-      The Metal Toolchain NAR must be available.
-
-      If you have cachix configured for exo.cachix.org, this should be automatic.
-
-      Otherwise:
-        1. Install Xcode 26+ from the App Store
-        2. Run: xcodebuild -downloadComponent MetalToolchain
-        3. Export the toolchain:
-           hdiutil attach "$(find /System/Library/AssetsV2/com_apple_MobileAsset_MetalToolchain -name '*.dmg' | head -1)" -mountpoint /tmp/metal-dmg
-           cp -R /tmp/metal-dmg/Metal.xctoolchain /tmp/metal-export
-           hdiutil detach /tmp/metal-dmg
-        4. Create NAR and add to store:
-           nix nar pack /tmp/metal-export > /tmp/metal-toolchain-17C48.nar
-           nix store add --mode flat /tmp/metal-toolchain-17C48.nar
-    '';
-    hash = "sha256-ayR5mXN4sZAddwKEG2OszGRF93k9ZFc7H0yi2xbylQw=";
-  };
-in
-stdenvNoCC.mkDerivation {
-  pname = "metal-toolchain";
-  version = "17C48";
-
-  dontUnpack = true;
-  dontBuild = true;
-  dontFixup = true;
-
-  nativeBuildInputs = [ nix ];
-
-  installPhase = ''
-    runHook preInstall
-
-    nix-store --restore $out < ${narFile}
-
-    # Create bin directory with symlinks for PATH
-    mkdir -p $out/bin
-    ln -s $out/usr/bin/metal $out/bin/metal
-    ln -s $out/usr/bin/metallib $out/bin/metallib
-
-    runHook postInstall
-  '';
-
-  # Metal language version for CMake (from: echo __METAL_VERSION__ | metal -E -x metal -P -)
-  passthru.metalVersion = "400";
-
-  meta = {
-    description = "Apple Metal compiler toolchain";
-    platforms = [ "aarch64-darwin" ];
-    license = lib.licenses.unfree;
-  };
-}

nix/mlx.nix

@@ -1,158 +0,0 @@
-{ stdenv
-, lib
-, fetchFromGitHub
-, replaceVars
-, fetchzip
-, cmake
-, nlohmann_json
-, apple-sdk_26
-, metal-toolchain
-, runCommand
-, fmt
-, python313Packages
-, uvLockMlxVersion
-}:
-
-assert stdenv.isDarwin;
-
-let
-  python = python313Packages.python;
-
-  # Static dependencies included directly during compilation
-  gguf-tools = fetchFromGitHub {
-    owner = "antirez";
-    repo = "gguf-tools";
-    rev = "8fa6eb65236618e28fd7710a0fba565f7faa1848";
-    hash = "sha256-15FvyPOFqTOr5vdWQoPnZz+mYH919++EtghjozDlnSA=";
-  };
-
-  metal_cpp = fetchzip {
-    url = "https://developer.apple.com/metal/cpp/files/metal-cpp_26.zip";
-    hash = "sha256-7n2eI2lw/S+Us6l7YPAATKwcIbRRpaQ8VmES7S8ZjY8=";
-  };
-
-  nanobind = fetchFromGitHub {
-    owner = "wjakob";
-    repo = "nanobind";
-    rev = "v2.10.2";
-    hash = "sha256-io44YhN+VpfHFWyvvLWSanRgbzA0whK8WlDNRi3hahU=";
-    fetchSubmodules = true;
-  };
-
-  mlx = stdenv.mkDerivation rec {
-    pname = "mlx";
-    version = let v = "0.30.4"; in
-      assert v == uvLockMlxVersion || throw "MLX version mismatch: nix/mlx.nix has ${v} but uv.lock has ${uvLockMlxVersion}. Update both the version and hash in nix/mlx.nix.";
-      v;
-    pyproject = true;
-
-    src = fetchFromGitHub {
-      owner = "ml-explore";
-      repo = "mlx";
-      tag = "v${version}";
-      hash = "sha256-OJk6jPlbaSlsUdk3ADz3tWcRzTWXRof3/q8Soe1AO6w=";
-    };
-
-    patches = [
-      (replaceVars ./darwin-build-fixes.patch {
-        sdkVersion = apple-sdk_26.version;
-        metalVersion = metal-toolchain.metalVersion;
-      })
-    ];
-
-    postPatch = ''
-      substituteInPlace mlx/backend/cpu/jit_compiler.cpp \
-        --replace-fail "g++" "$CXX"
-    '';
-
-    dontUseCmakeConfigure = true;
-
-    enableParallelBuilding = true;
-
-    # Allows multiple cores to be used in Python builds.
-    postUnpack = ''
-      export MAKEFLAGS+="''${enableParallelBuilding:+-j$NIX_BUILD_CORES}"
-    '';
-
-    # Updates the wrong fetcher rev attribute
-    passthru.skipBulkUpdate = true;
-
-    env = {
-      DEV_RELEASE = 1;
-      CMAKE_ARGS = toString [
-        (lib.cmakeBool "USE_SYSTEM_FMT" true)
-        (lib.cmakeOptionType "filepath" "FETCHCONTENT_SOURCE_DIR_GGUFLIB" "${gguf-tools}")
-        (lib.cmakeOptionType "filepath" "FETCHCONTENT_SOURCE_DIR_JSON" "${nlohmann_json.src}")
-        (lib.cmakeOptionType "filepath" "FETCHCONTENT_SOURCE_DIR_NANOBIND" "${nanobind}")
-        (lib.cmakeBool "FETCHCONTENT_FULLY_DISCONNECTED" true)
-        (lib.cmakeBool "MLX_BUILD_METAL" true)
-        (lib.cmakeOptionType "filepath" "FETCHCONTENT_SOURCE_DIR_METAL_CPP" "${metal_cpp}")
-        (lib.cmakeOptionType "string" "CMAKE_OSX_DEPLOYMENT_TARGET" "${apple-sdk_26.version}")
-        (lib.cmakeOptionType "filepath" "CMAKE_OSX_SYSROOT" "${apple-sdk_26.passthru.sdkroot}")
-      ];
-      SDKROOT = apple-sdk_26.passthru.sdkroot;
-      MACOSX_DEPLOYMENT_TARGET = apple-sdk_26.version;
-    };
-
-    build-system = [
-      python313Packages.setuptools
-    ];
-
-    nativeBuildInputs = [
-      cmake
-      metal-toolchain
-      python313Packages.pypaBuildHook
-      python313Packages.pypaInstallHook
-      python313Packages.setuptools
-      python313Packages.typing-extensions
-      python313Packages.wheel
-      python313Packages.cmake
-      python313Packages.ninja
-    ];
-
-    buildInputs = [
-      fmt
-      gguf-tools
-      python313Packages.nanobind
-      python313Packages.pybind11
-      apple-sdk_26
-    ];
-
-    # Tests require Metal GPU access which isn't available in the Nix sandbox.
-    # To run tests, build with: nix build --option sandbox false .#mlx.passthru.tests.mlxTest
-    doCheck = false;
-
-    pythonImportsCheck = [ "mlx" ];
-
-    passthru.tests = {
-      # Runs example scripts to verify MLX works. Requires --option sandbox false
-      # since Metal GPU access is needed.
-      mlxTest =
-        runCommand "run-mlx-examples"
-          {
-            buildInputs = [ mlx ];
-            nativeBuildInputs = [ python ];
-          }
-          ''
-            cp ${src}/examples/python/logistic_regression.py .
-            ${python.interpreter} logistic_regression.py
-            rm logistic_regression.py
-
-            cp ${src}/examples/python/linear_regression.py .
-            ${python.interpreter} linear_regression.py
-            rm linear_regression.py
-
-            touch $out
-          '';
-    };
-
-    meta = {
-      homepage = "https://github.com/ml-explore/mlx";
-      description = "Array framework for Apple silicon";
-      changelog = "https://github.com/ml-explore/mlx/releases/tag/${src.tag}";
-      license = lib.licenses.mit;
-      platforms = [ "aarch64-darwin" ];
-    };
-  };
-in
-mlx

pyproject.toml

@@ -17,8 +17,8 @@ dependencies = [
     "loguru>=0.7.3",
     "exo_pyo3_bindings", # rust bindings
     "anyio==4.11.0",
-    "mlx==0.30.4; sys_platform == 'darwin'",
-    "mlx[cpu]==0.30.4; sys_platform == 'linux'",
+    "mlx @ git+https://github.com/rltakashige/mlx-jaccl-fix-small-recv.git; sys_platform == 'darwin'",
+    "mlx[cpu]==0.30.3; sys_platform == 'linux'",
     "mlx-lm",
     "tiktoken>=0.12.0", # required for kimi k2 tokenizer
     "hypercorn>=0.18.0",

python/parts.nix

@@ -1,93 +0,0 @@
-{ inputs, ... }:
-{
-  perSystem =
-    { config, self', pkgs, lib, system, ... }:
-    let
-      # Load workspace from uv.lock
-      workspace = inputs.uv2nix.lib.workspace.loadWorkspace {
-        workspaceRoot = inputs.self;
-      };
-
-      # Create overlay from workspace
-      # Use wheels from PyPI for most packages; we override mlx with our pure Nix Metal build
-      overlay = workspace.mkPyprojectOverlay { sourcePreference = "wheel"; };
-
-      # Override overlay to inject Nix-built components
-      exoOverlay = final: prev: {
-        # Replace workspace exo_pyo3_bindings with Nix-built wheel
-        exo-pyo3-bindings = pkgs.stdenv.mkDerivation {
-          pname = "exo-pyo3-bindings";
-          version = "0.1.0";
-          src = self'.packages.exo_pyo3_bindings;
-          # Install from pre-built wheel
-          nativeBuildInputs = [ final.pyprojectWheelHook ];
-          dontStrip = true;
-        };
-      };
-
-      python = pkgs.python313;
-
-      # Overlay to provide build systems and custom packages
-      buildSystemsOverlay = final: prev: {
-        # Use our pure Nix-built MLX with Metal support
-        mlx = self'.packages.mlx;
-
-        # mlx-lm is a git dependency that needs setuptools
-        mlx-lm = prev.mlx-lm.overrideAttrs (old: {
-          nativeBuildInputs = (old.nativeBuildInputs or [ ]) ++ [
-            final.setuptools
-          ];
-        });
-      };
-
-      pythonSet = (pkgs.callPackage inputs.pyproject-nix.build.packages {
-        inherit python;
-      }).overrideScope (
-        lib.composeManyExtensions [
-          inputs.pyproject-build-systems.overlays.default
-          overlay
-          exoOverlay
-          buildSystemsOverlay
-        ]
-      );
-      exoVenv = pythonSet.mkVirtualEnv "exo-env" workspace.deps.default;
-
-      # Virtual environment with dev dependencies for testing
-      testVenv = pythonSet.mkVirtualEnv "exo-test-env" (
-        workspace.deps.default // {
-          exo = [ "dev" ]; # Include pytest, pytest-asyncio, pytest-env
-        }
-      );
-
-      exoPackage = pkgs.runCommand "exo"
-        {
-          nativeBuildInputs = [ pkgs.makeWrapper ];
-        }
-        ''
-          mkdir -p $out/bin
-
-          # Create wrapper scripts
-          for script in exo exo-master exo-worker; do
-            makeWrapper ${exoVenv}/bin/$script $out/bin/$script \
-              --set DASHBOARD_DIR ${self'.packages.dashboard}
-          done
-        '';
-    in
-    {
-      # Python package only available on macOS (requires MLX/Metal)
-      packages = lib.optionalAttrs pkgs.stdenv.hostPlatform.isDarwin {
-        exo = exoPackage;
-        # Test environment for running pytest outside of Nix sandbox (needs GPU access)
-        exo-test-env = testVenv;
-      };
-
-      checks = {
-        # Ruff linting (works on all platforms)
-        lint = pkgs.runCommand "ruff-lint" { } ''
-          export RUFF_CACHE_DIR="$TMPDIR/ruff-cache"
-          ${pkgs.ruff}/bin/ruff check ${inputs.self}/
-          touch $out
-        '';
-      };
-    };
-}

src/exo/download/impl_shard_downloader.py

@@ -21,7 +21,7 @@ def exo_shard_downloader(max_parallel_downloads: int = 8) -> ShardDownloader:
 
 
 async def build_base_shard(model_id: ModelId) -> ShardMetadata:
-    model_card = await ModelCard.load(model_id)
+    model_card = await ModelCard.from_hf(model_id)
     return PipelineShardMetadata(
         model_card=model_card,
         device_rank=0,
@@ -166,8 +166,9 @@ class ResumableShardDownloader(ShardDownloader):
         for task in asyncio.as_completed(tasks):
             try:
                 yield await task
+            # TODO: except Exception
             except Exception as e:
-                logger.warning(f"Error downloading shard: {type(e).__name__}")
+                logger.error("Error downloading shard:", e)
 
     async def get_shard_download_status_for_shard(
         self, shard: ShardMetadata

src/exo/master/api.py

@@ -65,9 +65,7 @@ from exo.shared.types.api import (
     StartDownloadParams,
     StartDownloadResponse,
     StreamingChoiceResponse,
-    StreamOptions,
     ToolCall,
-    Usage,
 )
 from exo.shared.types.chunks import (
     ErrorChunk,
@@ -115,9 +113,7 @@ def _format_to_content_type(image_format: Literal["png", "jpeg", "webp"] | None)
 
 
 def chunk_to_response(
-    chunk: TokenChunk | ToolCallChunk,
-    command_id: CommandId,
-    usage: Usage | None,
+    chunk: TokenChunk | ToolCallChunk, command_id: CommandId
 ) -> ChatCompletionResponse:
     return ChatCompletionResponse(
         id=command_id,
@@ -142,10 +138,21 @@ def chunk_to_response(
                 finish_reason=chunk.finish_reason,
             )
         ],
-        usage=usage,
     )
 
 
+async def resolve_model_card(model_id: ModelId) -> ModelCard:
+    if model_id in MODEL_CARDS:
+        model_card = MODEL_CARDS[model_id]
+        return model_card
+
+    for card in MODEL_CARDS.values():
+        if card.model_id == ModelId(model_id):
+            return card
+
+    return await ModelCard.from_hf(model_id)
+
+
 class API:
     def __init__(
         self,
@@ -267,7 +274,7 @@ class API:
 
     async def place_instance(self, payload: PlaceInstanceParams):
         command = PlaceInstance(
-            model_card=await ModelCard.load(payload.model_id),
+            model_card=await resolve_model_card(payload.model_id),
             sharding=payload.sharding,
             instance_meta=payload.instance_meta,
             min_nodes=payload.min_nodes,
@@ -284,7 +291,7 @@ class API:
         self, payload: CreateInstanceParams
     ) -> CreateInstanceResponse:
         instance = payload.instance
-        model_card = await ModelCard.load(instance.shard_assignments.model_id)
+        model_card = await resolve_model_card(instance.shard_assignments.model_id)
         required_memory = model_card.storage_size
         available_memory = self._calculate_total_available_memory()
 
@@ -312,7 +319,7 @@ class API:
         instance_meta: InstanceMeta = InstanceMeta.MlxRing,
         min_nodes: int = 1,
     ) -> Instance:
-        model_card = await ModelCard.load(model_id)
+        model_card = await resolve_model_card(model_id)
 
         try:
             placements = get_instance_placements(
@@ -515,10 +522,9 @@ class API:
                 del self._chat_completion_queues[command_id]
 
     async def _generate_chat_stream(
-        self, command_id: CommandId, stream_options: StreamOptions | None = None
+        self, command_id: CommandId
     ) -> AsyncGenerator[str, None]:
         """Generate chat completion stream as JSON strings."""
-        include_usage = stream_options.include_usage if stream_options else False
 
         async for chunk in self._chat_chunk_stream(command_id):
             assert not isinstance(chunk, ImageChunk)
@@ -534,10 +540,8 @@ class API:
                 yield "data: [DONE]\n\n"
                 return
 
-            usage = chunk.usage if include_usage else None
-
             chunk_response: ChatCompletionResponse = chunk_to_response(
-                chunk, command_id, usage=usage
+                chunk, command_id
             )
             logger.debug(f"chunk_response: {chunk_response}")
 
@@ -553,9 +557,8 @@ class API:
 
         text_parts: list[str] = []
         tool_calls: list[ToolCall] = []
-        model: ModelId | None = None
+        model: str | None = None
         finish_reason: FinishReason | None = None
-        usage: Usage | None = None
 
         async for chunk in self._chat_chunk_stream(command_id):
             if isinstance(chunk, ErrorChunk):
@@ -580,9 +583,6 @@ class API:
                     for i, tool in enumerate(chunk.tool_calls)
                 )
 
-            if chunk.usage is not None:
-                usage = chunk.usage
-
             if chunk.finish_reason is not None:
                 finish_reason = chunk.finish_reason
 
@@ -604,7 +604,6 @@ class API:
                     finish_reason=finish_reason,
                 )
             ],
-            usage=usage,
         )
 
     async def _collect_chat_completion_with_stats(
@@ -612,7 +611,7 @@ class API:
     ) -> BenchChatCompletionResponse:
         text_parts: list[str] = []
         tool_calls: list[ToolCall] = []
-        model: ModelId | None = None
+        model: str | None = None
         finish_reason: FinishReason | None = None
 
         stats: GenerationStats | None = None
@@ -665,7 +664,7 @@ class API:
         )
         return resp
 
-    async def _trigger_notify_user_to_download_model(self, model_id: ModelId) -> None:
+    async def _trigger_notify_user_to_download_model(self, model_id: str) -> None:
         logger.warning(
             "TODO: we should send a notification to the user to download the model"
         )
@@ -674,7 +673,7 @@ class API:
         self, payload: ChatCompletionTaskParams
     ) -> ChatCompletionResponse | StreamingResponse:
         """Handle chat completions, supporting both streaming and non-streaming responses."""
-        model_card = await ModelCard.load(ModelId(payload.model))
+        model_card = await resolve_model_card(ModelId(payload.model))
         payload.model = model_card.model_id
 
         if not any(
@@ -692,7 +691,7 @@ class API:
         await self._send(command)
         if payload.stream:
             return StreamingResponse(
-                self._generate_chat_stream(command.command_id, payload.stream_options),
+                self._generate_chat_stream(command.command_id),
                 media_type="text/event-stream",
             )
 
@@ -701,7 +700,7 @@ class API:
     async def bench_chat_completions(
         self, payload: BenchChatCompletionTaskParams
     ) -> BenchChatCompletionResponse:
-        model_card = await ModelCard.load(ModelId(payload.model))
+        model_card = await resolve_model_card(ModelId(payload.model))
         payload.model = model_card.model_id
 
         if not any(
@@ -721,12 +720,12 @@ class API:
         response = await self._collect_chat_completion_with_stats(command.command_id)
         return response
 
-    async def _validate_image_model(self, model: ModelId) -> ModelId:
+    async def _validate_image_model(self, model: str) -> ModelId:
         """Validate model exists and return resolved model ID.
 
         Raises HTTPException 404 if no instance is found for the model.
         """
-        model_card = await ModelCard.load(model)
+        model_card = await resolve_model_card(ModelId(model))
         resolved_model = model_card.model_id
         if not any(
             instance.shard_assignments.model_id == resolved_model
@@ -772,7 +771,7 @@ 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.model = await self._validate_image_model(payload.model)
 
         command = ImageGeneration(
             request_params=payload,
@@ -1017,7 +1016,7 @@ class API:
     async def bench_image_generations(
         self, request: Request, payload: BenchImageGenerationTaskParams
     ) -> BenchImageGenerationResponse:
-        payload.model = await self._validate_image_model(ModelId(payload.model))
+        payload.model = await self._validate_image_model(payload.model)
 
         payload.stream = False
         payload.partial_images = 0
@@ -1038,7 +1037,7 @@ class API:
         self,
         image: UploadFile,
         prompt: str,
-        model: ModelId,
+        model: str,
         n: int,
         size: str,
         response_format: Literal["url", "b64_json"],
@@ -1133,7 +1132,7 @@ class API:
         command = await self._send_image_edits_command(
             image=image,
             prompt=prompt,
-            model=ModelId(model),
+            model=model,
             n=n,
             size=size,
             response_format=response_format,
@@ -1189,7 +1188,7 @@ class API:
         command = await self._send_image_edits_command(
             image=image,
             prompt=prompt,
-            model=ModelId(model),
+            model=model,
             n=n,
             size=size,
             response_format=response_format,

src/exo/routing/router.py

@@ -216,8 +216,6 @@ def get_node_id_keypair(
     Obtains the :class:`Keypair` associated with this node-ID.
     Obtain the :class:`PeerId` by from it.
     """
-    # TODO(evan): bring back node id persistence once we figure out how to deal with duplicates
-    return Keypair.generate_ed25519()
 
     def lock_path(path: str | bytes | PathLike[str] | PathLike[bytes]) -> Path:
         return Path(str(path) + ".lock")

src/exo/shared/models/model_cards.py

@@ -1,5 +1,5 @@
 from enum import Enum
-from typing import Annotated, Any
+from typing import Annotated
 
 import aiofiles
 import aiofiles.os as aios
@@ -7,14 +7,7 @@ import tomlkit
 from anyio import Path, open_file
 from huggingface_hub import model_info
 from loguru import logger
-from pydantic import (
-    AliasChoices,
-    BaseModel,
-    Field,
-    PositiveInt,
-    field_validator,
-    model_validator,
-)
+from pydantic import BaseModel, Field, PositiveInt, field_validator
 
 from exo.shared.constants import EXO_ENABLE_IMAGE_MODELS
 from exo.shared.types.common import ModelId
@@ -718,18 +711,15 @@ if EXO_ENABLE_IMAGE_MODELS:
 class ConfigData(BaseModel):
     model_config = {"extra": "ignore"}  # Allow unknown fields
 
-    architectures: list[str] | None = None
+    # Common field names for number of layers across different architectures
+    num_hidden_layers: Annotated[int, Field(ge=0)] | None = None
+    num_layers: Annotated[int, Field(ge=0)] | None = None
+    n_layer: Annotated[int, Field(ge=0)] | None = None
+    n_layers: Annotated[int, Field(ge=0)] | None = None  # Sometimes used
+    num_decoder_layers: Annotated[int, Field(ge=0)] | None = None  # Transformer models
+    decoder_layers: Annotated[int, Field(ge=0)] | None = None  # Some architectures
     hidden_size: Annotated[int, Field(ge=0)] | None = None
-    layer_count: int = Field(
-        validation_alias=AliasChoices(
-            "num_hidden_layers",
-            "num_layers",
-            "n_layer",
-            "n_layers",
-            "num_decoder_layers",
-            "decoder_layers",
-        )
-    )
+    architectures: list[str] | None = None
 
     @property
     def supports_tensor(self) -> bool:
@@ -744,27 +734,25 @@ class ConfigData(BaseModel):
             ["GptOssForCausalLM"],
         ]
 
-    @model_validator(mode="before")
-    @classmethod
-    def defer_to_text_config(cls, data: dict[str, Any]):
-        text_config = data.get("text_config")
-        if text_config is None:
-            return data
+    @property
+    def layer_count(self) -> int:
+        # Check common field names for layer count
+        layer_fields = [
+            self.num_hidden_layers,
+            self.num_layers,
+            self.n_layer,
+            self.n_layers,
+            self.num_decoder_layers,
+            self.decoder_layers,
+        ]
 
-        for field in [
-            "architectures",
-            "hidden_size",
-            "num_hidden_layers",
-            "num_layers",
-            "n_layer",
-            "n_layers",
-            "num_decoder_layers",
-            "decoder_layers",
-        ]:
-            if (val := text_config.get(field)) is not None:  # pyright: ignore[reportAny]
-                data[field] = val
+        for layer_count in layer_fields:
+            if layer_count is not None:
+                return layer_count
 
-        return data
+        raise ValueError(
+            f"No layer count found in config.json: {self.model_dump_json()}"
+        )
 
 
 async def get_config_data(model_id: ModelId) -> ConfigData:

src/exo/shared/tests/test_node_id_persistence.py

@@ -8,7 +8,7 @@ from multiprocessing.synchronize import Event as EventT
 from multiprocessing.synchronize import Semaphore as SemaphoreT
 
 from loguru import logger
-from pytest import LogCaptureFixture, mark
+from pytest import LogCaptureFixture
 
 from exo.routing.router import get_node_id_keypair
 from exo.shared.constants import EXO_NODE_ID_KEYPAIR
@@ -74,7 +74,6 @@ def _delete_if_exists(p: str | bytes | os.PathLike[str] | os.PathLike[bytes]):
         os.remove(p)
 
 
-@mark.skip(reason="this functionality is currently disabled but may return in future")
 def test_node_id_fetching(caplog: LogCaptureFixture):
     reps = 10
 

src/exo/shared/types/api.py

@@ -11,7 +11,7 @@ from exo.shared.types.common import CommandId, NodeId
 from exo.shared.types.memory import Memory
 from exo.shared.types.worker.instances import Instance, InstanceId, InstanceMeta
 from exo.shared.types.worker.shards import Sharding, ShardMetadata
-from exo.utils.pydantic_ext import CamelCaseModel, ConfigDict, TaggedModel
+from exo.utils.pydantic_ext import CamelCaseModel
 
 FinishReason = Literal[
     "stop", "length", "tool_calls", "content_filter", "function_call", "error"
@@ -116,8 +116,8 @@ class Usage(BaseModel):
     prompt_tokens: int
     completion_tokens: int
     total_tokens: int
-    prompt_tokens_details: PromptTokensDetails
-    completion_tokens_details: CompletionTokensDetails
+    prompt_tokens_details: PromptTokensDetails | None = None
+    completion_tokens_details: CompletionTokensDetails | None = None
 
 
 class StreamingChoiceResponse(BaseModel):
@@ -170,13 +170,7 @@ class BenchChatCompletionResponse(ChatCompletionResponse):
     generation_stats: GenerationStats | None = None
 
 
-class StreamOptions(BaseModel):
-    include_usage: bool = False
-
-
-class ChatCompletionTaskParams(TaggedModel):
-    model_config = ConfigDict(extra="ignore")
-
+class ChatCompletionTaskParams(BaseModel):
     model: str
     frequency_penalty: float | None = None
     messages: list[ChatCompletionMessage]
@@ -190,7 +184,6 @@ class ChatCompletionTaskParams(TaggedModel):
     seed: int | None = None
     stop: str | list[str] | None = None
     stream: bool = False
-    stream_options: StreamOptions | None = None
     temperature: float | None = None
     top_p: float | None = None
     tools: list[dict[str, Any]] | None = None

src/exo/shared/types/chunks.py

@@ -2,7 +2,7 @@ from collections.abc import Generator
 from typing import Any, Literal
 
 from exo.shared.models.model_cards import ModelId
-from exo.shared.types.api import GenerationStats, ImageGenerationStats, Usage
+from exo.shared.types.api import GenerationStats, ImageGenerationStats
 from exo.utils.pydantic_ext import TaggedModel
 
 from .api import FinishReason
@@ -17,7 +17,6 @@ class BaseChunk(TaggedModel):
 class TokenChunk(BaseChunk):
     text: str
     token_id: int
-    usage: Usage | None
     finish_reason: Literal["stop", "length", "content_filter"] | None = None
     stats: GenerationStats | None = None
 
@@ -29,7 +28,6 @@ class ErrorChunk(BaseChunk):
 
 class ToolCallChunk(BaseChunk):
     tool_calls: list[ToolCallItem]
-    usage: Usage | None
     finish_reason: Literal["tool_calls"] = "tool_calls"
     stats: GenerationStats | None = None
 

src/exo/shared/types/commands.py

@@ -2,7 +2,6 @@ from pydantic import Field
 
 from exo.shared.models.model_cards import ModelCard, ModelId
 from exo.shared.types.api import (
-    BenchChatCompletionTaskParams,
     ChatCompletionTaskParams,
     ImageEditsInternalParams,
     ImageGenerationTaskParams,
@@ -23,7 +22,7 @@ class TestCommand(BaseCommand):
 
 
 class ChatCompletion(BaseCommand):
-    request_params: ChatCompletionTaskParams | BenchChatCompletionTaskParams
+    request_params: ChatCompletionTaskParams
 
 
 class ImageGeneration(BaseCommand):

src/exo/shared/types/tasks.py

@@ -3,7 +3,6 @@ from enum import Enum
 from pydantic import Field
 
 from exo.shared.types.api import (
-    BenchChatCompletionTaskParams,
     ChatCompletionTaskParams,
     ImageEditsInternalParams,
     ImageGenerationTaskParams,
@@ -55,7 +54,7 @@ class StartWarmup(BaseTask):  # emitted by Worker
 
 class ChatCompletion(BaseTask):  # emitted by Master
     command_id: CommandId
-    task_params: ChatCompletionTaskParams | BenchChatCompletionTaskParams
+    task_params: ChatCompletionTaskParams
 
     error_type: str | None = Field(default=None)
     error_message: str | None = Field(default=None)

src/exo/shared/types/worker/runner_response.py

@@ -6,7 +6,6 @@ from exo.shared.types.api import (
     GenerationStats,
     ImageGenerationStats,
     ToolCallItem,
-    Usage,
 )
 from exo.utils.pydantic_ext import TaggedModel
 
@@ -25,7 +24,6 @@ class GenerationResponse(BaseRunnerResponse):
     # logprobs: list[float] | None = None # too big. we can change to be top-k
     finish_reason: FinishReason | None = None
     stats: GenerationStats | None = None
-    usage: Usage | None
 
 
 class ImageGenerationResponse(BaseRunnerResponse):
@@ -59,7 +57,6 @@ class PartialImageResponse(BaseRunnerResponse):
 
 class ToolCallResponse(BaseRunnerResponse):
     tool_calls: list[ToolCallItem]
-    usage: Usage | None
 
 
 class FinishedResponse(BaseRunnerResponse):

src/exo/worker/engines/image/generate.py

@@ -98,8 +98,8 @@ def generate_image(
 
     partial_images = (
         task.partial_images
-        if task.partial_images is not None and task.stream is not None and task.stream
-        else 0
+        if task.partial_images is not None
+        else (3 if task.stream else 0)
     )
 
     image_path: Path | None = None

src/exo/worker/engines/image/pipeline/runner.py

@@ -348,7 +348,6 @@ class DiffusionRunner:
                 ctx.in_loop(  # pyright: ignore[reportAny]
                     t=t,
                     latents=latents,
-                    time_steps=time_steps,
                 )
 
                 mx.eval(latents)

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

@@ -201,9 +201,6 @@ def pipeline_auto_parallel(
     device_rank, world_size = model_shard_meta.device_rank, model_shard_meta.world_size
 
     layers = layers[start_layer:end_layer]
-    for layer in layers:
-        mx.eval(layer)  # type: ignore
-
     layers[0] = PipelineFirstLayer(layers[0], device_rank, group=group)
     layers[-1] = PipelineLastLayer(
         layers[-1],

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

@@ -3,7 +3,6 @@ from copy import deepcopy
 from typing import Any, cast
 
 import mlx.core as mx
-import psutil
 from mlx_lm.models.cache import (
     KVCache,
     QuantizedKVCache,
@@ -13,29 +12,25 @@ from mlx_lm.models.cache import (
 from mlx_lm.models.gpt_oss import Model as GptOssModel
 from mlx_lm.tokenizer_utils import TokenizerWrapper
 
-from exo.shared.types.memory import Memory
 from exo.shared.types.mlx import KVCacheType
 from exo.worker.engines.mlx import Model
 from exo.worker.engines.mlx.constants import CACHE_GROUP_SIZE, KV_CACHE_BITS
 from exo.worker.runner.bootstrap import logger
 
 # Fraction of device memory above which LRU eviction kicks in
-_DEFAULT_MEMORY_THRESHOLD = 0.9
+_DEFAULT_MEMORY_THRESHOLD = 0.85
 _MEMORY_THRESHOLD = float(
     os.environ.get("EXO_MEMORY_THRESHOLD", _DEFAULT_MEMORY_THRESHOLD)
 )
 
 
 class KVPrefixCache:
-    def __init__(
-        self, tokenizer: TokenizerWrapper, group: mx.distributed.Group | None = None
-    ):
+    def __init__(self, tokenizer: TokenizerWrapper):
         self.prompts: list[mx.array] = []  # mx array of tokens (ints)
         self.caches: list[KVCacheType] = []
         self._last_used: list[int] = []  # monotonic counter of last access per entry
         self._access_counter: int = 0
         self._tokenizer: TokenizerWrapper = tokenizer
-        self._group = group
 
     def clear(self):
         """Clear all cached prompts and caches."""
@@ -86,13 +81,13 @@ class KVPrefixCache:
         best_snapshot_index, best_snapshot_length = None, 0
 
         for i, cached_prompt in enumerate(self.prompts):
-            length = get_prefix_length(tokenized_prompt, cached_prompt)
+            length = _get_prefix_length(tokenized_prompt, cached_prompt)
 
             if length == max_length:
                 # Exact match - cached prompt starts with our entire prompt
                 # Trim cache to prompt length - 1, return last token for stream_generate
                 prompt_cache = deepcopy(self.caches[i])
-                cached_length = cache_length(self.caches[i])
+                cached_length = _cache_length(self.caches[i])
                 tokens_to_trim = cached_length - (max_length - 1)
                 if tokens_to_trim > 0:
                     trim_prompt_cache(cast(list[Any], prompt_cache), tokens_to_trim)
@@ -114,7 +109,7 @@ class KVPrefixCache:
             prompt_cache = deepcopy(self.caches[best_snapshot_index])
 
             # Trim removes tokens from the end, so we trim (cached_length - prefix_length) to keep the prefix
-            cached_length = cache_length(self.caches[best_snapshot_index])
+            cached_length = _cache_length(self.caches[best_snapshot_index])
             tokens_to_trim = cached_length - best_snapshot_length
             if tokens_to_trim > 0:
                 trim_prompt_cache(cast(list[Any], prompt_cache), tokens_to_trim)
@@ -136,37 +131,29 @@ class KVPrefixCache:
             return prompt_cache, tokenized_prompt, None
 
     def _evict_if_needed(self):
-        """Evict least recently used entries while memory usage is high."""
+        """Evict least recently used entries while memory pressure is high."""
         if len(self.caches) == 0:
             return
 
+        active: int = mx.metal.get_active_memory()
+        limit = int(mx.metal.device_info()["max_recommended_working_set_size"])
+        if active < limit * _MEMORY_THRESHOLD:
+            return
+
         # Evict LRU entries until below threshold or only one entry left
-        while (
-            len(self.caches) > 1
-            and self.get_memory_used_percentage() > _MEMORY_THRESHOLD
-        ):
+        while len(self.caches) > 0:
             lru_index = self._last_used.index(min(self._last_used))
             evicted_tokens = len(self.prompts[lru_index])
             self.prompts.pop(lru_index)
             self.caches.pop(lru_index)
             self._last_used.pop(lru_index)
             logger.info(
-                f"KV cache evicted LRU entry ({evicted_tokens} tokens) due to memory usage"
+                f"KV cache evicted LRU entry ({evicted_tokens} tokens) due to memory pressure"
             )
 
-    def get_memory_used_percentage(self) -> float:
-        local_pressure: float = get_memory_used_percentage()
-
-        if self._group is None:
-            return local_pressure
-
-        all_pressure = mx.distributed.all_gather(
-            mx.array([local_pressure], dtype=mx.float32),
-            group=self._group,
-        )
-        # .item() evals.
-        max_pressure = float(mx.max(all_pressure).item())
-        return max_pressure
+            active = mx.metal.get_active_memory()
+            if active < limit * _MEMORY_THRESHOLD:
+                break
 
 
 def encode_prompt(tokenizer: TokenizerWrapper, prompt: str) -> mx.array:
@@ -181,13 +168,13 @@ def encode_prompt(tokenizer: TokenizerWrapper, prompt: str) -> mx.array:
     return mx.array(tokenized_prompt)
 
 
-def cache_length(cache: KVCacheType) -> int:
+def _cache_length(cache: KVCacheType) -> int:
     """Get the number of tokens in a KV cache."""
     # Use .offset attribute which all cache types have (len() not implemented in older QuantizedKVCache)
     return max(c.offset for c in cache)  # type: ignore
 
 
-def get_prefix_length(prompt: mx.array, cached_prompt: mx.array) -> int:
+def _get_prefix_length(prompt: mx.array, cached_prompt: mx.array) -> int:
     """Find the length of the common prefix between two token arrays."""
     n = min(int(prompt.shape[0]), int(cached_prompt.shape[0]))
     if n == 0:
@@ -198,17 +185,6 @@ def get_prefix_length(prompt: mx.array, cached_prompt: mx.array) -> int:
     return int(mx.sum(prefix_mask).item())
 
 
-def get_available_memory() -> Memory:
-    mem: int = psutil.virtual_memory().available
-    return Memory.from_bytes(mem)
-
-
-def get_memory_used_percentage() -> float:
-    mem = psutil.virtual_memory()
-    # percent is 0-100
-    return float(mem.percent / 100)
-
-
 def make_kv_cache(
     model: Model, max_kv_size: int | None = None, keep: int = 0
 ) -> KVCacheType:

src/exo/worker/engines/mlx/generator/generate.py

@@ -10,11 +10,8 @@ from mlx_lm.tokenizer_utils import TokenizerWrapper
 from exo.shared.types.api import (
     BenchChatCompletionTaskParams,
     ChatCompletionMessage,
-    CompletionTokensDetails,
     FinishReason,
     GenerationStats,
-    PromptTokensDetails,
-    Usage,
 )
 from exo.shared.types.memory import Memory
 from exo.shared.types.mlx import KVCacheType
@@ -42,7 +39,7 @@ def prefill(
     sampler: Callable[[mx.array], mx.array],
     prompt_tokens: mx.array,
     cache: KVCacheType,
-) -> tuple[float, int]:
+) -> float:
     """Prefill the KV cache with prompt tokens.
 
     This runs the model over the prompt tokens to populate the cache,
@@ -53,7 +50,7 @@ def prefill(
     """
     num_tokens = len(prompt_tokens)
     if num_tokens == 0:
-        return 0.0, 0
+        return 0.0
 
     logger.debug(f"Prefilling {num_tokens} tokens...")
     start_time = time.perf_counter()
@@ -88,7 +85,7 @@ def prefill(
         f"Prefill complete: {num_tokens} tokens in {elapsed:.2f}s "
         f"({tokens_per_sec:.1f} tok/s)"
     )
-    return tokens_per_sec, num_tokens
+    return tokens_per_sec
 
 
 def warmup_inference(
@@ -172,8 +169,6 @@ def mlx_generate(
     mx.reset_peak_memory()
     is_bench: bool = isinstance(task, BenchChatCompletionTaskParams)
 
-    logger.info(f"{is_bench=}")
-
     # Currently we support chat-completion tasks only.
     logger.debug(f"task_params: {task}")
 
@@ -209,9 +204,7 @@ def mlx_generate(
     )
 
     # Prefill cache with all tokens except the last one
-    prefill_tps, prefill_tokens = prefill(
-        model, tokenizer, sampler, prompt_tokens[:-1], caches
-    )
+    prefill_tps = prefill(model, tokenizer, sampler, prompt_tokens[:-1], caches)
 
     # stream_generate starts from the last token
     last_token = prompt_tokens[-1:]
@@ -219,43 +212,28 @@ def mlx_generate(
     max_tokens = task.max_tokens or MAX_TOKENS
     generated_text_parts: list[str] = []
     generation_start_time = time.perf_counter()
-    usage: Usage | None = None
-    in_thinking = False
-    reasoning_tokens = 0
-    think_start = tokenizer.think_start
-    think_end = tokenizer.think_end
-    for completion_tokens, out in enumerate(
-        stream_generate(
-            model=model,
-            tokenizer=tokenizer,
-            prompt=last_token,
-            max_tokens=max_tokens,
-            sampler=sampler,
-            logits_processors=logits_processors,
-            prompt_cache=caches,
-            # TODO: Dynamically change prefill step size to be the maximum possible without timing out.
-            prefill_step_size=2048,
-            kv_group_size=KV_GROUP_SIZE,
-            kv_bits=KV_BITS,
-        ),
-        start=1,
+    for out in stream_generate(
+        model=model,
+        tokenizer=tokenizer,
+        prompt=last_token,
+        max_tokens=max_tokens,
+        sampler=sampler,
+        logits_processors=logits_processors,
+        prompt_cache=caches,
+        # TODO: Dynamically change prefill step size to be the maximum possible without timing out.
+        prefill_step_size=2048,
+        kv_group_size=KV_GROUP_SIZE,
+        kv_bits=KV_BITS,
     ):
         generated_text_parts.append(out.text)
         logger.info(out.text)
 
-        if think_start is not None and out.text == think_start:
-            in_thinking = True
-        elif think_end is not None and out.text == think_end:
-            in_thinking = False
-        if in_thinking:
-            reasoning_tokens += 1
-
         stats: GenerationStats | None = None
         if out.finish_reason is not None:
             stats = GenerationStats(
                 prompt_tps=float(prefill_tps or out.prompt_tps),
                 generation_tps=float(out.generation_tps),
-                prompt_tokens=int(prefill_tokens + out.prompt_tokens),
+                prompt_tokens=int(out.prompt_tokens),
                 generation_tokens=int(out.generation_tokens),
                 peak_memory_usage=Memory.from_gb(out.peak_memory),
             )
@@ -267,24 +245,11 @@ def mlx_generate(
                     f"Model generated unexpected finish_reason: {out.finish_reason}"
                 )
 
-            usage = Usage(
-                prompt_tokens=int(out.prompt_tokens),
-                completion_tokens=completion_tokens,
-                total_tokens=int(out.prompt_tokens) + completion_tokens,
-                prompt_tokens_details=PromptTokensDetails(
-                    cached_tokens=prefix_hit_length
-                ),
-                completion_tokens_details=CompletionTokensDetails(
-                    reasoning_tokens=reasoning_tokens
-                ),
-            )
-
         yield GenerationResponse(
             text=out.text,
             token=out.token,
             finish_reason=cast(FinishReason | None, out.finish_reason),
             stats=stats,
-            usage=usage,
         )
 
         if out.finish_reason is not None:

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

@@ -165,6 +165,7 @@ def mlx_distributed_init(
 
                 jaccl_coordinator = jaccl_coordinators[bound_instance.bound_node_id]
 
+                # TODO: update once upstream fixes
                 logger.info(
                     f"rank {rank} MLX_IBV_DEVICES: {coordination_file} with devices: {jaccl_devices_json}"
                 )

src/exo/worker/runner/runner.py

@@ -163,7 +163,7 @@ def main(
                         logger.info(
                             f"model has_tool_calling={tokenizer.has_tool_calling}"
                         )
-                        kv_prefix_cache = KVPrefixCache(tokenizer, group)
+                        kv_prefix_cache = KVPrefixCache(tokenizer)
 
                     elif (
                         ModelTask.TextToImage in shard_metadata.model_card.tasks
@@ -277,11 +277,9 @@ def main(
                                 tokenizer.tool_parser,  # pyright: ignore[reportAny]
                             )
 
-                        completion_tokens = 0
                         for response in mlx_generator:
                             match response:
                                 case GenerationResponse():
-                                    completion_tokens += 1
                                     if (
                                         device_rank == 0
                                         and response.finish_reason == "error"
@@ -309,7 +307,6 @@ def main(
                                                     model=shard_metadata.model_card.model_id,
                                                     text=response.text,
                                                     token_id=response.token,
-                                                    usage=response.usage,
                                                     finish_reason=response.finish_reason,
                                                     stats=response.stats,
                                                 ),
@@ -323,7 +320,6 @@ def main(
                                                 chunk=ToolCallChunk(
                                                     tool_calls=response.tool_calls,
                                                     model=shard_metadata.model_card.model_id,
-                                                    usage=response.usage,
                                                 ),
                                             )
                                         )
@@ -539,10 +535,10 @@ def parse_gpt_oss(
                             name=current_tool_name,
                             arguments="".join(tool_arg_parts).strip(),
                         )
-                    ],
-                    usage=response.usage,
+                    ]
                 )
                 tool_arg_parts = []
+                break
             current_tool_name = recipient
 
         # If inside a tool call, accumulate arguments
@@ -688,7 +684,7 @@ def parse_tool_calls(
                     tools = [_validate_single_tool(tool) for tool in parsed]
                 else:
                     tools = [_validate_single_tool(parsed)]
-                yield ToolCallResponse(tool_calls=tools, usage=response.usage)
+                yield ToolCallResponse(tool_calls=tools)
 
             except (
                 json.JSONDecodeError,

src/exo/worker/tests/unittests/test_mlx/test_kv_prefix_cache.py

@@ -14,9 +14,9 @@ from exo.shared.types.tasks import ChatCompletionTaskParams
 from exo.worker.engines.mlx import Model
 from exo.worker.engines.mlx.cache import (
     KVPrefixCache,
-    cache_length,
+    _cache_length,
+    _get_prefix_length,
     encode_prompt,
-    get_prefix_length,
     make_kv_cache,
 )
 from exo.worker.engines.mlx.generator.generate import mlx_generate, prefill
@@ -35,47 +35,47 @@ class TestGetPrefixLength:
     def test_identical_arrays(self):
         a = mx.array([1, 2, 3, 4, 5])
         b = mx.array([1, 2, 3, 4, 5])
-        assert get_prefix_length(a, b) == 5
+        assert _get_prefix_length(a, b) == 5
 
     def test_no_common_prefix(self):
         a = mx.array([1, 2, 3])
         b = mx.array([4, 5, 6])
-        assert get_prefix_length(a, b) == 0
+        assert _get_prefix_length(a, b) == 0
 
     def test_partial_prefix(self):
         a = mx.array([1, 2, 3, 4, 5])
         b = mx.array([1, 2, 3, 7, 8])
-        assert get_prefix_length(a, b) == 3
+        assert _get_prefix_length(a, b) == 3
 
     def test_prompt_longer_than_cached(self):
         a = mx.array([1, 2, 3, 4, 5])
         b = mx.array([1, 2, 3])
-        assert get_prefix_length(a, b) == 3
+        assert _get_prefix_length(a, b) == 3
 
     def test_cached_longer_than_prompt(self):
         a = mx.array([1, 2, 3])
         b = mx.array([1, 2, 3, 4, 5])
-        assert get_prefix_length(a, b) == 3
+        assert _get_prefix_length(a, b) == 3
 
     def test_single_token_match(self):
         a = mx.array([1, 2, 3])
         b = mx.array([1, 5, 6])
-        assert get_prefix_length(a, b) == 1
+        assert _get_prefix_length(a, b) == 1
 
     def test_empty_prompt(self):
         a = mx.array([]).astype(mx.int32)
         b = mx.array([1, 2, 3])
-        assert get_prefix_length(a, b) == 0
+        assert _get_prefix_length(a, b) == 0
 
     def test_empty_cached(self):
         a = mx.array([1, 2, 3])
         b = mx.array([]).astype(mx.int32)
-        assert get_prefix_length(a, b) == 0
+        assert _get_prefix_length(a, b) == 0
 
     def test_both_empty(self):
         a = mx.array([]).astype(mx.int32)
         b = mx.array([]).astype(mx.int32)
-        assert get_prefix_length(a, b) == 0
+        assert _get_prefix_length(a, b) == 0
 
 
 class TestKVPrefix:
@@ -146,7 +146,7 @@ class TestKVPrefixCacheWithModel:
         prefill(model, tokenizer, make_sampler(0.0), tokens, cache)
 
         # Cache should now hold the prompt tokens
-        assert cache_length(cache) == len(tokens)
+        assert _cache_length(cache) == len(tokens)
 
     def test_add_and_get_exact_match(self, model_and_tokenizer):
         model, tokenizer = model_and_tokenizer
@@ -166,7 +166,7 @@ class TestKVPrefixCacheWithModel:
         kv_prefix_cache.add_kv_cache(prompt, cache)
 
         assert len(kv_prefix_cache.prompts) == 1
-        stored_length = cache_length(kv_prefix_cache.caches[0])
+        stored_length = _cache_length(kv_prefix_cache.caches[0])
         assert stored_length > 0
 
         # Retrieve with same prompt: exact match
@@ -209,7 +209,7 @@ class TestKVPrefixCacheWithModel:
         long_tokens = encode_prompt(tokenizer, long_prompt)
 
         # The prompts share a prefix (chat template preamble + "Hi")
-        expected_prefix = get_prefix_length(long_tokens, short_tokens)
+        expected_prefix = _get_prefix_length(long_tokens, short_tokens)
         assert expected_prefix > 0, (
             "Prompts should share a prefix from the chat template"
         )
@@ -243,7 +243,7 @@ class TestKVPrefixCacheWithModel:
         kv_prefix_cache = KVPrefixCache(tokenizer)
         kv_prefix_cache.add_kv_cache(prompt, cache)
 
-        stored_length = cache_length(kv_prefix_cache.caches[0])
+        stored_length = _cache_length(kv_prefix_cache.caches[0])
 
         # Get cache and mutate it (simulating what generation does)
         result_cache, _, matched_index = kv_prefix_cache.get_kv_cache(model, prompt)
@@ -259,7 +259,7 @@ class TestKVPrefixCacheWithModel:
         mx.eval([c.keys for c in result_cache])
 
         # Stored cache must be unchanged
-        assert cache_length(kv_prefix_cache.caches[0]) == stored_length
+        assert _cache_length(kv_prefix_cache.caches[0]) == stored_length
 
     def test_stored_cache_survives_repeated_get_mutate_cycles(
         self, model_and_tokenizer
@@ -281,7 +281,7 @@ class TestKVPrefixCacheWithModel:
         kv_prefix_cache = KVPrefixCache(tokenizer)
         kv_prefix_cache.add_kv_cache(prompt, cache)
 
-        stored_length = cache_length(kv_prefix_cache.caches[0])
+        stored_length = _cache_length(kv_prefix_cache.caches[0])
 
         for i in range(3):
             result_cache, _, _ = kv_prefix_cache.get_kv_cache(model, prompt)
@@ -293,7 +293,7 @@ class TestKVPrefixCacheWithModel:
                 layer_cache.update_and_fetch(extra, extra)
             mx.eval([c.keys for c in result_cache])
 
-            assert cache_length(kv_prefix_cache.caches[0]) == stored_length, (
+            assert _cache_length(kv_prefix_cache.caches[0]) == stored_length, (
                 f"Failed on loop {i}"
             )
 
@@ -325,7 +325,7 @@ class TestKVPrefixCacheWithModel:
         assert len(kv_prefix_cache.caches) == 1
         # Cache should contain prompt + generated tokens
         expected_length = len(prompt_tokens) + generated_tokens
-        assert cache_length(kv_prefix_cache.caches[0]) == expected_length
+        assert _cache_length(kv_prefix_cache.caches[0]) == expected_length
 
     def test_mlx_generate_second_call_gets_prefix_hit(self, model_and_tokenizer):
         """Second mlx_generate call with same prompt should get a prefix hit from stored cache."""
@@ -400,7 +400,7 @@ class TestKVPrefixCacheWithModel:
         first_gen_time = time.perf_counter() - t0
 
         assert len(kv_prefix_cache.prompts) == 1
-        first_cache_length = cache_length(kv_prefix_cache.caches[0])
+        first_cache_length = _cache_length(kv_prefix_cache.caches[0])
 
         # Second generation: same long prompt + extra content (simulating multi-turn)
         task2 = ChatCompletionTaskParams(
@@ -416,7 +416,7 @@ class TestKVPrefixCacheWithModel:
         prompt2_tokens = encode_prompt(tokenizer, prompt2)
 
         # Verify the prompts share a long prefix
-        prefix_len = get_prefix_length(prompt2_tokens, prompt1_tokens)
+        prefix_len = _get_prefix_length(prompt2_tokens, prompt1_tokens)
         assert prefix_len > 1000, "Prompts must share > 1000 token prefix"
 
         # Second generation should reuse the cached prefix (only prefill new tokens)
@@ -440,7 +440,7 @@ class TestKVPrefixCacheWithModel:
         # With prefix_hit > 1000, should update in-place (not add a second entry)
         assert len(kv_prefix_cache.prompts) == 1
         # Updated cache should be longer (prompt2 + generated > prompt1 + generated)
-        updated_cache_length = cache_length(kv_prefix_cache.caches[0])
+        updated_cache_length = _cache_length(kv_prefix_cache.caches[0])
         assert updated_cache_length > first_cache_length
 
     def test_mlx_generate_stored_cache_not_mutated(self, model_and_tokenizer):
@@ -465,7 +465,7 @@ class TestKVPrefixCacheWithModel:
         ):
             pass
 
-        firstcache_length = cache_length(kv_prefix_cache.caches[0])
+        first_cache_length = _cache_length(kv_prefix_cache.caches[0])
 
         # Second generation gets the cache and mutates it during generation
         for _response in mlx_generate(
@@ -478,7 +478,7 @@ class TestKVPrefixCacheWithModel:
             pass
 
         # The first stored cache must not have been mutated by the second generation
-        assert cache_length(kv_prefix_cache.caches[0]) == firstcache_length
+        assert _cache_length(kv_prefix_cache.caches[0]) == first_cache_length
 
     def test_evicts_lru_entry_under_memory_pressure(self, model_and_tokenizer):
         """Under memory pressure, adding a new cache entry evicts the least recently used one."""
@@ -540,6 +540,6 @@ class TestKVPrefixCacheWithModel:
         assert len(kv_prefix_cache.prompts) == 1
         # The surviving entry should be the newly added one
         new_tokens = encode_prompt(tokenizer, prompt)
-        assert get_prefix_length(kv_prefix_cache.prompts[0], new_tokens) == len(
+        assert _get_prefix_length(kv_prefix_cache.prompts[0], new_tokens) == len(
             new_tokens
         )

src/exo/worker/tests/unittests/test_runner/test_event_ordering.py

@@ -109,8 +109,8 @@ def assert_events_equal(test_events: Iterable[Event], true_events: Iterable[Even
 
 @pytest.fixture
 def patch_out_mlx(monkeypatch: pytest.MonkeyPatch):
-    # initialize_mlx returns a mock group
-    monkeypatch.setattr(mlx_runner, "initialize_mlx", make_nothin(MockGroup()))
+    # initialize_mlx returns a "group" equal to 1
+    monkeypatch.setattr(mlx_runner, "initialize_mlx", make_nothin(1))
     monkeypatch.setattr(mlx_runner, "load_mlx_items", make_nothin((1, MockTokenizer)))
     monkeypatch.setattr(mlx_runner, "warmup_inference", make_nothin(1))
     monkeypatch.setattr(mlx_runner, "_check_for_debug_prompts", nothin)
@@ -120,7 +120,7 @@ def patch_out_mlx(monkeypatch: pytest.MonkeyPatch):
     monkeypatch.setattr(mlx_runner, "detect_thinking_prompt_suffix", make_nothin(False))
 
     def fake_generate(*_1: object, **_2: object):
-        yield GenerationResponse(token=0, text="hi", finish_reason="stop", usage=None)
+        yield GenerationResponse(token=0, text="hi", finish_reason="stop")
 
     monkeypatch.setattr(mlx_runner, "mlx_generate", fake_generate)
 
@@ -147,14 +147,6 @@ class MockTokenizer:
     has_tool_calling = False
 
 
-class MockGroup:
-    def rank(self) -> int:
-        return 0
-
-    def size(self) -> int:
-        return 1
-
-
 def _run(tasks: Iterable[Task]):
     bound_instance = get_bound_mlx_ring_instance(
         instance_id=INSTANCE_1_ID,
@@ -190,8 +182,6 @@ def test_events_processed_in_correct_order(patch_out_mlx: pytest.MonkeyPatch):
             text="hi",
             token_id=0,
             finish_reason="stop",
-            usage=None,
-            stats=None,
         ),
     )
 

tests/start_distributed_test.sh

@@ -11,6 +11,7 @@ if [[ $# -lt 2 ]]; then
   exit 1
 fi
 
+
 kind=$1
 shift
 
@@ -30,14 +31,14 @@ for name in "${hostnames[@]}"; do
   weaved+=("$name" "$ip")
 done
 
-devs_raw=$(printf '["%s", "%s"], ' "${weaved[@]}")
+devs_raw=$(printf "[\"%s\", \"%s\"], " "${weaved[@]}")
 devs="[${devs_raw%, }]"
 
 model_ids=("qwen3-30b" "gpt-oss-120b-MXFP4-Q8" "kimi-k2-thinking")
 
 for model_id in "${model_ids[@]}"; do
-  for i in "${!ips[@]}"; do
-    {
+  for i in "${!ips[@]}"; do  
+    { 
       req="{
         \"model_id\": \"${model_id}\",
         \"devs\": ${devs},
@@ -47,8 +48,9 @@ for model_id in "${model_ids[@]}"; do
       curl -sN \
         -X POST "http://${ips[$i]}:52415/${kind}" \
         -H "Content-Type: application/json" -d "$req" \
-        2>&1 | sed "s/^/\n${hostnames[$i]}@${ips[$i]}: /" || echo "curl to ${hostnames[$i]} failed" && exit 1
+      2>&1 | sed "s/^/\n${hostnames[$i]}@${ips[$i]}: /" || echo "curl to ${hostnames[$i]} failed" && exit 1
     } &
   done
   wait
 done
+

tmp/config_examples/opencode.json

@@ -1,18 +0,0 @@
-{
-  "$schema": "https://opencode.ai/config.json",
-  "model": "exo/mlx-community/gpt-oss-120b-MXFP4-Q8",
-  "provider": {
-    "exo": {
-      "api": "http://localhost:52415/v1",
-      "models": {
-        "mlx-community/gpt-oss-120b-MXFP4-Q8": {
-          "name": "GPT OSS 120B",
-          "limit": {
-            "context": 32768,
-            "output": 8192
-          }
-        }
-      }
-    }
-  }
-}

tmp/set_rdma_network_config.sh

@@ -1,47 +0,0 @@
-#!/usr/bin/env bash
-
-set -euo pipefail
-
-PREFS="/Library/Preferences/SystemConfiguration/preferences.plist"
-
-# Remove bridge0 interface
-ifconfig bridge0 &>/dev/null && {
-  ifconfig bridge0 | grep -q 'member' && {
-    ifconfig bridge0 | awk '/member/ {print $2}' | xargs -n1 ifconfig bridge0 deletem 2>/dev/null || true
-  }
-  ifconfig bridge0 destroy 2>/dev/null || true
-}
-
-# Remove Thunderbolt Bridge from VirtualNetworkInterfaces in preferences.plist
-/usr/libexec/PlistBuddy -c "Delete :VirtualNetworkInterfaces:Bridge:bridge0" "$PREFS" 2>/dev/null || true
-
-networksetup -listlocations | grep -q exo || {
-  networksetup -createlocation exo
-}
-
-networksetup -switchtolocation exo
-networksetup -listallhardwareports |
-  awk -F': ' '/Hardware Port: / {print $2}' |
-  while IFS=":" read -r name; do
-    case "$name" in
-    "Ethernet Adapter"*) ;;
-    "Thunderbolt Bridge") ;;
-    "Thunderbolt "*)
-      networksetup -listallnetworkservices |
-        grep -q "EXO $name" ||
-        networksetup -createnetworkservice "EXO $name" "$name" 2>/dev/null ||
-        continue
-      networksetup -setdhcp "EXO $name"
-      ;;
-    *)
-      networksetup -listallnetworkservices |
-        grep -q "$name" ||
-        networksetup -createnetworkservice "$name" "$name" 2>/dev/null ||
-        continue
-      ;;
-    esac
-  done
-
-networksetup -listnetworkservices | grep -q "Thunderbolt Bridge" && {
-  networksetup -setnetworkserviceenabled "Thunderbolt Bridge" off
-} || true