@oonacode/win32-x64 0.3.271 → 0.3.273

package/oona.exe

@@ -1,4 +1,4 @@
 [diffend] Oversized file quarantined before diffing.
 name: package/oona.exe
-size: 113249280 bytes
-sha256: 8bf75de543976f4010f0ee907b20ce485a8adc0d365962e6a18faabb97d261f6
+size: 113326080 bytes
+sha256: 65acf7beade73702bdaf4d15099952fc0b715972059c86a8f1f640f4b3065ffd

package/oonacode-release.json

@@ -1,13 +1,13 @@
 {
     "name":  "oonacode",
-    "version":  "0.3.271",
+    "version":  "0.3.273",
     "channel":  "stable",
     "platform":  "win32",
     "arch":  "x64",
     "mode":  "bun",
     "launcherRuntime":  "bun 1.4.1",
-    "commit":  "d6cea4ca",
-    "builtAt":  "2026-10-25T17:19:27.3847600Z",
+    "commit":  "558bb1c8",
+    "builtAt":  "2026-10-26T17:39:58.7169242Z",
     "engine":  "engine-store",
     "engineVersion":  "0.3.301",
     "siteOrigin":  "https://oonacode.oonak.ai",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@oonacode/win32-x64",
-  "version": "0.3.271",
+  "version": "0.3.273",
   "description": "OonaCode native binaries for Windows x64: oona.exe (CLI + headless host service), the pinned oonaclaude.exe engine sidecar, local web UI assets. Installed automatically as an optional dependency of `oonacode`.",
   "license": "UNLICENSED",
   "homepage": "https://oonacode.oonak.ai",

package/registry.json

@@ -1781,21 +1781,16 @@
       "models": [
         {
           "id": "nvidia/nemotron-3-ultra-550b-a55b",
+          "helperModel": "nvidia/nemotron-3-ultra-550b-a55b",
           "displayName": "Nemotron 3 Ultra 550B (NIM)",
           "contextWindow": 262144,
           "supportsTools": true,
           "supportsThinking": true,
           "supportsVision": false
         },
-        {
-          "id": "nvidia/nemotron-3-super-120b-a12b",
-          "displayName": "Nemotron 3 Super 120B (NIM)",
-          "supportsTools": true,
-          "supportsThinking": true,
-          "supportsVision": false
-        },
         {
           "id": "nvidia/nemotron-3.5-super-vl-120b-a12b",
+          "helperModel": "nvidia/nemotron-3.5-super-vl-120b-a12b",
           "displayName": "Nemotron 3.5 Super VL Intelligence (NIM)",
           "contextWindow": 262144,
           "supportsTools": true,
@@ -1805,6 +1800,7 @@
         },
         {
           "id": "nvidia/nemotron-3.5-lightning-30b-a3b",
+          "helperModel": "nvidia/nemotron-3.5-lightning-30b-a3b",
           "displayName": "Nemotron 3.5 Lightning 30B (NIM)",
           "supportsTools": true,
           "supportsThinking": true,
@@ -1812,72 +1808,34 @@
         },
         {
           "id": "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning",
+          "helperModel": "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning",
           "displayName": "Nemotron 3 Nano Omni 30B (NIM)",
           "contextWindow": 262144,
           "supportsTools": true,
           "supportsThinking": true,
           "supportsVision": true
         },
-        {
-          "id": "moonshotai/kimi-k3",
-          "displayName": "Kimi K3 (NIM)",
-          "contextWindow": 1048576,
-          "supportsTools": true,
-          "supportsThinking": true,
-          "supportsVision": true,
-          "notes": "NVIDIA's card says a multi-turn tool conversation must send back the earlier reasoning_content; the translator drops thinking blocks on replay, so long tool loops may degrade. Unverified."
-        },
-        {
-          "id": "z-ai/glm-5.3",
-          "displayName": "GLM-5.3 (NIM)",
-          "contextWindow": 1048576,
-          "supportsTools": true,
-          "supportsThinking": true,
-          "supportsVision": false
-        },
         {
           "id": "z-ai/glm-5.3-flash",
+          "helperModel": "z-ai/glm-5.3-flash",
           "displayName": "GLM-5.3 Flash (NIM)",
           "contextWindow": 1048576,
           "supportsTools": true,
           "supportsThinking": true,
           "supportsVision": true
         },
-        {
-          "id": "deepseek-ai/deepseek-v4.1-flash",
-          "displayName": "DeepSeek V4.1 Flash (NIM)",
-          "supportsTools": true,
-          "supportsThinking": true,
-          "supportsVision": false,
-          "notes": "NVIDIA's card was not readable (404), so no context window is set: the engine assumes 200k."
-        },
-        {
-          "id": "poolside/laguna-xs-2.1",
-          "displayName": "Poolside Laguna XS 2.1 (NIM)",
-          "contextWindow": 262144,
-          "supportsTools": true,
-          "supportsThinking": true,
-          "supportsVision": false
-        },
         {
           "id": "meta/muse-glimmer-30b",
+          "helperModel": "meta/muse-glimmer-30b",
           "displayName": "Meta Muse Glimmer 30B (NIM)",
           "contextWindow": 131072,
           "supportsTools": true,
           "supportsThinking": true,
           "supportsVision": true
         },
-        {
-          "id": "google/gemma-4-31b-it",
-          "displayName": "Gemma 4 31B (NIM)",
-          "contextWindow": 262144,
-          "supportsTools": true,
-          "supportsThinking": false,
-          "supportsVision": true,
-          "notes": "Thinking is switched by a token in the system prompt, which the translator does not send, so it runs without."
-        },
         {
           "id": "openai/gpt-oss-20b",
+          "helperModel": "openai/gpt-oss-20b",
           "displayName": "GPT-OSS 20B (NIM)",
           "contextWindow": 128000,
           "supportsTools": true,
@@ -1889,7 +1847,7 @@
       "modelAliases": {
         "opus": "nvidia/nemotron-3-ultra-550b-a55b",
         "sonnet": "nvidia/nemotron-3-ultra-550b-a55b",
-        "haiku": "nvidia/nemotron-3-super-120b-a12b"
+        "haiku": "nvidia/nemotron-3.5-lightning-30b-a3b"
       },
       "capabilities": {
         "streaming": true,
@@ -1910,7 +1868,7 @@
       "userBaseUrl": true,
       "phase": 1,
       "enabled": true,
-      "notes": "A user's OWN key only (owner, 2026-10-12: \"bring back the nvidia for local only, i have apikey want to use it now\"). The host translates it on the PC (engine/local-upstream.ts) and the key never reaches the gateway; the server never calls NVIDIA (SERVER_EXCLUDED_PROVIDER_IDS: no pool rows, no admin page). Models are NVIDIA's current hosted catalogue that call tools (build.nvidia.com, 2026-10-12); a context window is set only where the model card states one, otherwise the engine's 200k default applies. Left out: embeddings, safety filters, parsers, DiffusionGemma and the two Llama 3.2 vision models (no tool calling).",
+      "notes": "A user's OWN key (owner, 2026-10-12: \"bring back the nvidia for local only, i have apikey want to use it now\"). The host translates it on the PC (engine/local-upstream.ts) and the key never reaches the gateway. Seven models (owner, 2026-10-26: \"add only these\"), all NVIDIA-hosted and tool-calling (build.nvidia.com); a context window is set only where the model card states one, otherwise the engine's 200k default applies. Left out: embeddings, safety filters, parsers, and every model that timed out, answered 404 or reached end of life when measured with a real key.",
       "docsUrl": "https://docs.api.nvidia.com/nim/"
     }
   ]