pi-baseten-provider 1.0.6 → 1.0.8

package/README.md

@@ -71,15 +71,17 @@ pi
 |-------|---------|--------|-----------|-----------|------------|
 | DeepSeek V4 Pro | 131K | ❌ | ✅ | $1.74 | $3.48 |
 | GLM 4.7 | 200K | ❌ | ✅ | $0.60 | $2.20 |
-| GLM 5 | 203K | ❌ | ✅ | $0.95 | $3.15 |
-| GLM 5.1 | 203K | ❌ | ❌ | $1.30 | $4.30 |
-| GLM 5.2 | 203K | ❌ | ✅ | $1.40 | $4.40 |
-| Kimi K2.5 | 262K | ✅ | ✅ | $0.60 | $3.00 |
+| GLM 5.2 | 1.0M | ❌ | ✅ | $1.40 | $4.40 |
+| GLM 5.2 Fast | 524K | ❌ | ✅ | $2.10 | $6.60 |
+| Inkling | 1.0M | ❌ | ✅ | $1.00 | $4.05 |
+| Inkling Small | 1.0M | ❌ | ✅ | Free | Free |
 | Kimi K2.6 | 262K | ✅ | ✅ | $0.60 | $3.00 |
 | Kimi K2.7 Code | 262K | ❌ | ✅ | $0.95 | $4.00 |
-| Nemotron Super | 203K | ❌ | ✅ | $0.30 | $0.75 |
+| Kimi K3 | 1.0M | ❌ | ✅ | $3.00 | $15.00 |
+| Mercury 2 | 8K | ❌ | ✅ | Free | Free |
 | Nemotron Ultra | 203K | ❌ | ✅ | $0.60 | $2.40 |
 | OpenAI GPT 120B | 128K | ❌ | ✅ | $0.10 | $0.50 |
+| SID-1 | 33K | ❌ | ✅ | Free | Free |
 
 *Costs are per million tokens. Prices subject to change — check [baseten.co/pricing](https://www.baseten.co/pricing/) for current pricing.*
 

package/deprecated-models.json

@@ -0,0 +1 @@
+{}

package/index.ts

@@ -29,6 +29,7 @@ import { getAgentDir, type ExtensionAPI, type ModelRegistry } from "@earendil-wo
 import modelsData from "./models.json" with { type: "json" };
 import customModelsData from "./custom-models.json" with { type: "json" };
 import patchData from "./patch.json" with { type: "json" };
+import deprecatedData from "./deprecated-models.json" with { type: "json" };
 import fs from "fs";
 import path from "path";
 
@@ -38,7 +39,7 @@ interface JsonModel {
   id: string;
   name: string;
   reasoning: boolean;
-  input: string[];
+  input: ("text" | "image")[];
   cost: {
     input: number;
     output: number;
@@ -67,7 +68,7 @@ interface JsonModel {
 interface PatchEntry {
   name?: string;
   reasoning?: boolean;
-  input?: string[];
+  input?: ("text" | "image")[];
   cost?: {
     input?: number;
     output?: number;
@@ -123,7 +124,10 @@ function applyPatch(model: JsonModel, patch: PatchEntry): JsonModel {
 function buildModels(base: JsonModel[], custom: JsonModel[], patch: PatchData): JsonModel[] {
   const modelMap = new Map<string, JsonModel>();
 
-  for (const model of base) {
+  // Seed with the base list plus grace-period deprecated models so patch.json
+  // entries apply to deprecated models exactly as while the model was live
+  // (withDeprecated keeps live data on id conflicts).
+  for (const model of withDeprecated(base)) {
     modelMap.set(model.id, model);
   }
 
@@ -260,6 +264,35 @@ function mergeWithEmbedded(liveModels: JsonModel[], embeddedModels: JsonModel[])
   return result;
 }
 
+// Grace period for delisted models. When the provider API stops listing a
+// model, update-models.js moves its last-known definition into
+// deprecated-models.json (stamped with deprecatedAt) instead of dropping it.
+// For 14 days the model keeps working here so in-flight sessions and saved
+// model settings do not break; afterwards it is evicted permanently.
+const DEPRECATED_MODEL_TTL_MS = 14 * 24 * 60 * 60 * 1000;
+
+// Grace-period deprecated models with deprecation metadata stripped.
+function activeDeprecatedModels(): JsonModel[] {
+  const now = Date.now();
+  const result: JsonModel[] = [];
+  for (const entry of Object.values(deprecatedData as Record<string, JsonModel & { deprecatedAt?: string }>)) {
+    if (!entry?.id) continue;
+    const removedAt = Date.parse(entry.deprecatedAt ?? "");
+    if (Number.isNaN(removedAt) || now - removedAt > DEPRECATED_MODEL_TTL_MS) continue;
+    const model = { ...entry } as JsonModel & { deprecatedAt?: string };
+    delete model.deprecatedAt;
+    result.push(model);
+  }
+  return result;
+}
+
+// Append grace-period deprecated models the list does not already have (live data wins).
+function withDeprecated(models: JsonModel[]): JsonModel[] {
+  const seen = new Set(models.map((m) => m.id));
+  const extras = activeDeprecatedModels().filter((m) => !seen.has(m.id));
+  return extras.length > 0 ? [...models, ...extras] : models;
+}
+
 function loadStaleModels(embeddedModels: JsonModel[]): JsonModel[] {
   const cached = loadCachedModels();
   if (!cached || cached.length === 0) return embeddedModels;

package/models.json

@@ -44,20 +44,20 @@
     }
   },
   {
-    "id": "zai-org/GLM-5",
-    "name": "GLM 5",
-    "reasoning": false,
+    "id": "zai-org/GLM-5.2",
+    "name": "GLM 5.2",
+    "reasoning": true,
     "input": [
       "text"
     ],
     "cost": {
-      "input": 0.95,
-      "output": 3.15,
+      "input": 1.4,
+      "output": 4.4,
       "cacheRead": 0,
       "cacheWrite": 0
     },
-    "contextWindow": 202800,
-    "maxTokens": 202800,
+    "contextWindow": 1048576,
+    "maxTokens": 262144,
     "compat": {
       "supportsDeveloperRole": true,
       "supportsStore": false,
@@ -66,41 +66,42 @@
     }
   },
   {
-    "id": "zai-org/GLM-5.1",
-    "name": "GLM 5.1",
-    "reasoning": false,
+    "id": "zai-org/GLM-5.2-Fast",
+    "name": "GLM 5.2 Fast",
+    "reasoning": true,
     "input": [
       "text"
     ],
     "cost": {
-      "input": 1.3,
-      "output": 4.3,
+      "input": 2.1,
+      "output": 6.6,
       "cacheRead": 0,
       "cacheWrite": 0
     },
-    "contextWindow": 202800,
-    "maxTokens": 202800,
+    "contextWindow": 524288,
+    "maxTokens": 262144,
     "compat": {
       "supportsDeveloperRole": true,
       "supportsStore": false,
-      "maxTokensField": "max_completion_tokens"
+      "maxTokensField": "max_completion_tokens",
+      "thinkingFormat": "openai"
     }
   },
   {
-    "id": "zai-org/GLM-5.2",
-    "name": "GLM 5.2",
+    "id": "thinkingmachines/inkling",
+    "name": "Inkling",
     "reasoning": true,
     "input": [
       "text"
     ],
     "cost": {
-      "input": 1.4,
-      "output": 4.4,
+      "input": 1,
+      "output": 4.05,
       "cacheRead": 0,
       "cacheWrite": 0
     },
-    "contextWindow": 202720,
-    "maxTokens": 202720,
+    "contextWindow": 1048576,
+    "maxTokens": 32768,
     "compat": {
       "supportsDeveloperRole": true,
       "supportsStore": false,
@@ -109,21 +110,20 @@
     }
   },
   {
-    "id": "moonshotai/Kimi-K2.5",
-    "name": "Kimi K2.5",
-    "reasoning": false,
+    "id": "thinkingmachines/inkling-small",
+    "name": "Inkling Small",
+    "reasoning": true,
     "input": [
-      "text",
-      "image"
+      "text"
     ],
     "cost": {
-      "input": 0.6,
-      "output": 3,
+      "input": 0,
+      "output": 0,
       "cacheRead": 0,
       "cacheWrite": 0
     },
-    "contextWindow": 262000,
-    "maxTokens": 262000,
+    "contextWindow": 1048576,
+    "maxTokens": 32768,
     "compat": {
       "supportsDeveloperRole": true,
       "supportsStore": false,
@@ -177,20 +177,42 @@
     }
   },
   {
-    "id": "nvidia/Nemotron-120B-A12B",
-    "name": "Nemotron Super",
+    "id": "moonshotai/Kimi-K3",
+    "name": "Kimi K3",
     "reasoning": true,
     "input": [
       "text"
     ],
     "cost": {
-      "input": 0.3,
-      "output": 0.75,
+      "input": 3,
+      "output": 15,
       "cacheRead": 0,
       "cacheWrite": 0
     },
-    "contextWindow": 202800,
-    "maxTokens": 202800,
+    "contextWindow": 1048576,
+    "maxTokens": 262144,
+    "compat": {
+      "supportsDeveloperRole": true,
+      "supportsStore": false,
+      "maxTokensField": "max_completion_tokens",
+      "thinkingFormat": "openai"
+    }
+  },
+  {
+    "id": "inception/mercury-2",
+    "name": "Mercury 2",
+    "reasoning": true,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 8192,
+    "maxTokens": 5000,
     "compat": {
       "supportsDeveloperRole": true,
       "supportsStore": false,
@@ -242,5 +264,27 @@
       "thinkingFormat": "openai",
       "supportsReasoningEffort": true
     }
+  },
+  {
+    "id": "sid/sid-1",
+    "name": "SID-1",
+    "reasoning": true,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 32768,
+    "maxTokens": 5000,
+    "compat": {
+      "supportsDeveloperRole": true,
+      "supportsStore": false,
+      "maxTokensField": "max_completion_tokens",
+      "thinkingFormat": "openai"
+    }
   }
 ]

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-baseten-provider",
-  "version": "1.0.6",
+  "version": "1.0.8",
   "description": "Baseten provider extension for pi - Access DeepSeek, Kimi, GLM, MiniMax, Nemotron, and GPT-OSS models through the Baseten Model API",
   "author": "monotykamary",
   "homepage": "https://github.com/monotykamary/pi-baseten-provider#readme",
@@ -30,6 +30,7 @@
   "files": [
     "index.ts",
     "models.json",
+    "deprecated-models.json",
     "custom-models.json",
     "patch.json",
     "scripts/update-models.js"

package/scripts/update-models.js

@@ -276,6 +276,67 @@ function updateReadme(models) {
 
 // ─── Main ────────────────────────────────────────────────────────────────────
 
+// Grace period for delisted models: update-models.js moves models the API no
+// longer lists into deprecated-models.json (stamped with deprecatedAt) instead
+// of dropping them; the runtime appends them back so sessions and saved model
+// settings keep working, and after 14 days they are evicted permanently.
+const DEPRECATED_MODEL_TTL_MS = 14 * 24 * 60 * 60 * 1000;
+
+/**
+ * Reconcile deprecated-models.json against the freshly fetched model list.
+ * - in old models.json but not the API: moved into the deprecated file
+ *   (deprecatedAt = now; preserved on repeat runs so the grace clock is not reset)
+ * - back in the API: resurrected (dropped from the deprecated file)
+ * - deprecatedAt older than 14 days: evicted permanently
+ * Must run BEFORE the new models.json is written; it reads the old file itself.
+ */
+function updateDeprecatedModels(modelsJsonPath, newModels) {
+  const deprecatedPath = path.join(path.dirname(modelsJsonPath), 'deprecated-models.json');
+
+  let oldModels = [];
+  try {
+    const parsed = JSON.parse(fs.readFileSync(modelsJsonPath, 'utf8'));
+    if (Array.isArray(parsed)) oldModels = parsed;
+  } catch { /* first run: no previous models.json */ }
+
+  let deprecated = {};
+  try {
+    const parsed = JSON.parse(fs.readFileSync(deprecatedPath, 'utf8'));
+    if (parsed && typeof parsed === 'object' && !Array.isArray(parsed)) deprecated = parsed;
+  } catch { /* no graveyard yet */ }
+
+  const currentIds = new Set(newModels.map(m => m.id));
+  const now = new Date().toISOString();
+  const added = [];
+  const resurrected = [];
+  const evicted = [];
+
+  for (const old of oldModels) {
+    if (old && old.id && !currentIds.has(old.id) && !deprecated[old.id]) {
+      deprecated[old.id] = { ...old, deprecatedAt: now };
+      added.push(old.id);
+    }
+  }
+
+  for (const [id, entry] of Object.entries(deprecated)) {
+    if (currentIds.has(id)) {
+      delete deprecated[id];
+      resurrected.push(id);
+      continue;
+    }
+    const removedAt = Date.parse(entry && entry.deprecatedAt ? entry.deprecatedAt : '');
+    if (Number.isNaN(removedAt) || Date.now() - removedAt > DEPRECATED_MODEL_TTL_MS) {
+      delete deprecated[id];
+      evicted.push(id);
+    }
+  }
+
+  if (added.length > 0 || resurrected.length > 0 || evicted.length > 0) {
+    fs.writeFileSync(deprecatedPath, JSON.stringify(deprecated, null, 2) + '\n');
+    console.log('Updated deprecated-models.json ' + JSON.stringify({ added, resurrected, evicted }));
+  }
+}
+
 async function main() {
   try {
     const apiModels = await fetchModels();
@@ -299,6 +360,8 @@ async function main() {
     models.sort((a, b) => a.name.localeCompare(b.name));
 
     // Save models.json (pure API output, no patch/custom baked in)
+    // Move delisted models to deprecated-models.json BEFORE models.json is overwritten
+    updateDeprecatedModels(MODELS_JSON_PATH, models);
     saveJson(MODELS_JSON_PATH, models);
 
     // Build full model list for README: base → patch → custom