pi-runinfra-provider 1.0.0 → 1.2.0

package/README.md

@@ -15,7 +15,7 @@ _A [pi](https://github.com/earendil-works/pi-coding-agent) provider extension fo
 
 ## Features
 
-- **5 AI Models** including DeepSeek V4 Flash, DeepSeek V4 Pro, Nemotron 3.5 Lightning 30B, Qwen3.8 2.4T A95B, and Qwen3.8 27B
+- **10 AI Models** covering the full RunInfra lineup: DeepSeek V4 Flash, DeepSeek V4 Pro, Nemotron 3.5 Lightning 30B, Ornith 1.5 35B, Qwen3.8 2.4T A95B, Qwen3.8 27B, Parakeet TDT 0.6B v3, and the Qwen3 embedding/reranker models
 - **OpenAI-Compatible API** — Just change the base URL and API key
 - **Cost Tracking** — Per-model pricing for budget management
 - **Reasoning Models** — Extended thinking via `reasoning_content` field
@@ -71,7 +71,13 @@ pi
 |-------|---------|--------|-----------|-----------|-----------------|------------|
 | DeepSeek V4 Flash | 1.0M | ❌ | ✅ | $0.13 | $0.01 | $0.27 |
 | DeepSeek V4 Pro | 1.0M | ❌ | ✅ | $0.60 | $0.03 | $1.90 |
-| Nemotron 3.5 Lightning 30B | 262K | ❌ | ✅ | $0.05 | — | $0.15 |
+| GLM 5.3 Flash | 1.0M | ✅ | ✅ | $0.10 | $0.01 | $0.40 |
+| Nemotron 3.5 Lightning 30B | 262K | ❌ | ✅ | $0.05 | $0.01 | $0.15 |
+| Ornith 1.5 35B | 262K | ✅ | ✅ | $0.10 | $0.01 | $0.40 |
+| Parakeet TDT 0.6B v3 | 131K | ❌ | ❌ | $10.00 | — | — |
+| Qwen3 Embedding 0.6B | 33K | ❌ | ❌ | $0.01 | — | — |
+| Qwen3 Embedding 8B | 33K | ❌ | ❌ | $0.05 | — | — |
+| Qwen3 Reranker 8B | 16K | ❌ | ❌ | $0.05 | — | — |
 | Qwen3.8 2.4T A95B | 262K | ❌ | ✅ | $2.00 | $0.20 | $6.00 |
 | Qwen3.8 27B | 262K | ❌ | ✅ | $0.10 | $0.01 | $0.40 |
 
@@ -128,7 +134,7 @@ RunInfra's API follows the OpenAI Chat Completions API:
 - **`supportsDeveloperRole: false`** — All models. RunInfra uses standard OpenAI roles.
 - **`maxTokensField: "max_tokens"`** — All models. RunInfra accepts `max_tokens`.
 - **`thinkingFormat: "deepseek"`** — DeepSeek V4 Flash / V4 Pro. Reasoning is toggled via a `thinking` parameter and effort is sent as `reasoning_effort` (pi maps Shift+Tab levels to `low`/`high`/`max`).
-- **`thinkingFormat: "openai"`** — Nemotron 3.5 Lightning, Qwen3.8. Returns thinking in the `reasoning_content` field.
+- **`thinkingFormat: "openai"`** — Nemotron 3.5 Lightning, Ornith 1.5 35B, and Qwen3.8. Returns thinking in the `reasoning_content` field.
 - **`supportsReasoningEffort: true`** — All reasoning models. Supports the `reasoning_effort` parameter.
 - **`supportsStore: false`** — All models. RunInfra doesn't support the `store` parameter.
 
@@ -165,6 +171,13 @@ This will:
 3. Apply overrides from `patch.json`
 4. Update `models.json` and the README model table
 
+To regenerate just the README model table from local data — no API key needed,
+useful for offline curation:
+
+```bash
+node scripts/update-models.js --readme-only
+```
+
 ## License
 
 MIT

package/custom-models.json

@@ -1 +1,2 @@
 []
+

package/deprecated-models.json

@@ -1,24 +1,90 @@
 {
-  "nemotron-3-5-lightning-30b": {
-    "id": "nemotron-3-5-lightning-30b",
-    "name": "Nemotron 3.5 Lightning 30B",
-    "reasoning": true,
+  "parakeet-tdt-0-6b-v3": {
+    "id": "parakeet-tdt-0-6b-v3",
+    "name": "Parakeet TDT 0.6B v3",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 10,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 131072,
+    "maxTokens": 32768,
+    "compat": {
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "maxTokensField": "max_tokens"
+    },
+    "deprecatedAt": "2026-08-27T07:36:24.190Z"
+  },
+  "qwen3-embedding-0-6b": {
+    "id": "qwen3-embedding-0-6b",
+    "name": "Qwen3 Embedding 0.6B",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0.01,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 32768,
+    "maxTokens": 32768,
+    "compat": {
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "maxTokensField": "max_tokens"
+    },
+    "deprecatedAt": "2026-08-27T07:36:24.190Z"
+  },
+  "qwen3-embedding-8b": {
+    "id": "qwen3-embedding-8b",
+    "name": "Qwen3 Embedding 8B",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0.05,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 32768,
+    "maxTokens": 32768,
+    "compat": {
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "maxTokensField": "max_tokens"
+    },
+    "deprecatedAt": "2026-08-27T07:36:24.190Z"
+  },
+  "qwen3-reranker-8b": {
+    "id": "qwen3-reranker-8b",
+    "name": "Qwen3 Reranker 8B",
+    "reasoning": false,
     "input": [
       "text"
     ],
     "cost": {
       "input": 0.05,
-      "output": 0.15,
+      "output": 0,
       "cacheRead": 0,
       "cacheWrite": 0
     },
-    "contextWindow": 262144,
+    "contextWindow": 16384,
     "maxTokens": 32768,
     "compat": {
       "supportsDeveloperRole": false,
       "supportsStore": false,
       "maxTokensField": "max_tokens"
     },
-    "deprecatedAt": "2026-08-18T16:43:21.091Z"
+    "deprecatedAt": "2026-08-27T07:36:24.190Z"
   }
 }

package/models.json

@@ -13,7 +13,7 @@
       "cacheWrite": 0
     },
     "contextWindow": 1048576,
-    "maxTokens": 32768,
+    "maxTokens": 1048576,
     "compat": {
       "supportsDeveloperRole": false,
       "supportsStore": false,
@@ -34,7 +34,71 @@
       "cacheWrite": 0
     },
     "contextWindow": 1048576,
-    "maxTokens": 32768,
+    "maxTokens": 1048576,
+    "compat": {
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "maxTokensField": "max_tokens"
+    }
+  },
+  {
+    "id": "glm-5-3-flash",
+    "name": "Glm 5 3 Flash",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0.1,
+      "output": 0.4,
+      "cacheRead": 0.01,
+      "cacheWrite": 0
+    },
+    "contextWindow": 1048576,
+    "maxTokens": 1048576,
+    "compat": {
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "maxTokensField": "max_tokens"
+    }
+  },
+  {
+    "id": "nemotron-3-5-lightning-30b",
+    "name": "Nemotron 3.5 Lightning 30B",
+    "reasoning": true,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0.05,
+      "output": 0.15,
+      "cacheRead": 0.01,
+      "cacheWrite": 0
+    },
+    "contextWindow": 262144,
+    "maxTokens": 262144,
+    "compat": {
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "maxTokensField": "max_tokens"
+    }
+  },
+  {
+    "id": "ornith-1-5-35b",
+    "name": "Ornith 1.5 35B",
+    "reasoning": true,
+    "input": [
+      "text",
+      "image"
+    ],
+    "cost": {
+      "input": 0.1,
+      "output": 0.4,
+      "cacheRead": 0.01,
+      "cacheWrite": 0
+    },
+    "contextWindow": 262144,
+    "maxTokens": 262144,
     "compat": {
       "supportsDeveloperRole": false,
       "supportsStore": false,
@@ -55,7 +119,7 @@
       "cacheWrite": 0
     },
     "contextWindow": 262144,
-    "maxTokens": 32768,
+    "maxTokens": 262144,
     "compat": {
       "supportsDeveloperRole": false,
       "supportsStore": false,
@@ -76,7 +140,7 @@
       "cacheWrite": 0
     },
     "contextWindow": 262144,
-    "maxTokens": 32768,
+    "maxTokens": 262144,
     "compat": {
       "supportsDeveloperRole": false,
       "supportsStore": false,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-runinfra-provider",
-  "version": "1.0.0",
+  "version": "1.2.0",
   "description": "RunInfra provider extension for pi - Access DeepSeek V4, Nemotron 3.5 Lightning, and Qwen3.8 models through the RunInfra gateway",
   "author": "monotykamary",
   "homepage": "https://github.com/monotykamary/pi-runinfra-provider#readme",

package/patch.json

@@ -37,6 +37,19 @@
       "maxTokensField": "max_tokens"
     }
   },
+  "glm-5-3-flash": {
+    "name": "GLM 5.3 Flash",
+    "reasoning": true,
+    "input": [
+      "text",
+      "image"
+    ],
+    "compat": {
+      "thinkingFormat": "zai",
+      "supportsReasoningEffort": false,
+      "maxTokensField": "max_tokens"
+    }
+  },
   "nemotron-3-5-lightning-30b": {
     "name": "Nemotron 3.5 Lightning 30B",
     "reasoning": true,
@@ -56,6 +69,25 @@
       "maxTokensField": "max_tokens"
     }
   },
+  "ornith-1-5-35b": {
+    "reasoning": true,
+    "name": "Ornith 1.5 35B",
+    "thinkingLevelMap": {
+      "off": "none",
+      "minimal": "low",
+      "low": "low",
+      "medium": "medium",
+      "high": "xhigh",
+      "xhigh": "xhigh",
+      "max": "xhigh"
+    },
+    "compat": {
+      "thinkingFormat": "openai",
+      "supportsReasoningEffort": true,
+      "requiresReasoningContentOnAssistantMessages": true,
+      "maxTokensField": "max_tokens"
+    }
+  },
   "qwen3-8-2-4t-a95b": {
     "name": "Qwen3.8 2.4T A95B",
     "reasoning": true,

package/scripts/update-models.js

@@ -430,6 +430,20 @@ function withDeprecatedForReadme(models) {
 
 async function main() {
   try {
+    // Regenerate the derived README table from local source data without an API
+    // key. This is useful for offline workflows while curating custom models.
+    if (process.argv.includes('--readme-only')) {
+      const baseModels = loadJson(MODELS_JSON_PATH);
+      const patchData = loadJson(PATCH_JSON_PATH);
+      const customModels = loadJson(CUSTOM_MODELS_JSON_PATH);
+      const readmeBase = withDeprecatedForReadme(Array.isArray(baseModels) ? baseModels : []);
+      const readmeModels = buildModels(readmeBase, Array.isArray(customModels) ? customModels : [], patchData);
+      readmeModels.sort((a, b) => a.name.localeCompare(b.name));
+      updateReadme(readmeModels);
+      console.log('✓ Regenerated README.md from local model data');
+      return;
+    }
+
     const apiModels = await fetchModels();
 
     // Load existing models.json — source of truth for curated specs