@link-assistant/agent 0.26.11 → 0.26.13

package/README.md

@@ -20,11 +20,11 @@
 
 > This is the JavaScript/Bun implementation. See also the [Rust implementation](../rust/README.md).
 
-This is an MVP implementation of an OpenCode-compatible CLI agent, focused on maximum efficiency and unrestricted execution. We reproduce OpenCode's `run --format json --model opencode/minimax-m2.5-free` mode with:
+This is an MVP implementation of an OpenCode-compatible CLI agent, focused on maximum efficiency and unrestricted execution. We reproduce OpenCode's `run --format json` mode with:
 
-- ✅ **JSON Input/Output**: Compatible with `opencode run --format json --model opencode/minimax-m2.5-free`
+- ✅ **JSON Input/Output**: Compatible with `opencode run --format json`
 - ✅ **Plain Text Input**: Also accepts plain text messages (auto-converted to JSON format)
-- ✅ **Flexible Model Selection**: Defaults to free OpenCode Zen MiniMax M2.5 Free, supports [OpenCode Zen](https://opencode.ai/docs/zen/), [Formal AI](../docs/formal-ai.md), [Claude OAuth](../docs/claude-oauth.md), [Groq](../docs/groq.md), and [OpenRouter](../docs/openrouter.md) providers
+- ✅ **Flexible Model Selection**: Defaults to free Kilo Nemotron 3 Super (Kilo authentication recommended), supports [OpenCode Zen](https://opencode.ai/docs/zen/), [Formal AI](../docs/formal-ai.md), [Claude OAuth](../docs/claude-oauth.md), [Groq](../docs/groq.md), and [OpenRouter](../docs/openrouter.md) providers
 - ✅ **No Restrictions**: Fully unrestricted file system and command execution access (no sandbox)
 - ✅ **Minimal Footprint**: Built with Bun.sh for maximum efficiency
 - ✅ **Full Tool Support**: 13 tools including websearch, codesearch, batch - all enabled by default
@@ -169,7 +169,7 @@ echo '{"message":"hi"}' | agent
 **With custom model:**
 
 ```bash
-echo "hi" | agent --model opencode/minimax-m2.5-free
+echo "hi" | agent --model kilo/nemotron-3-super-free
 ```
 
 ### More Examples
@@ -190,16 +190,13 @@ echo '{"message":"run command","tools":[{"name":"bash","params":{"command":"ls -
 **Using different models:**
 
 ```bash
-# Default model (free MiniMax M2.5)
+# Default model (free Nemotron 3 Super)
 echo "hi" | agent
 
-# Other free models (in order of recommendation)
-echo "hi" | agent --model opencode/minimax-m2.5-free
-echo "hi" | agent --model opencode/ling-2.6-flash-free
-echo "hi" | agent --model opencode/hy3-preview-free
-echo "hi" | agent --model opencode/nemotron-3-super-free
-echo "hi" | agent --model opencode/gpt-5-nano
-echo "hi" | agent --model opencode/big-pickle
+# Built-in free models (authenticate with agent auth login or KILO_API_KEY)
+echo "hi" | agent --model kilo/nemotron-3-super-free
+echo "hi" | agent --model kilo/nemotron-3-ultra-free
+echo "hi" | agent --model kilo/nemotron-3.5-lightning-free
 
 # Formal AI local server
 echo "hi" | agent --model formal-ai
@@ -288,7 +285,7 @@ agent [options]
 
 Options:
   --model                        Model to use in format providerID/modelID
-                                 Default: opencode/minimax-m2.5-free
+                                 Default: kilo/nemotron-3-super-free
   --json-standard                JSON output format standard
                                  Choices: "opencode" (default), "claude" (experimental)
   --output-format                Alias for JSON output format

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@link-assistant/agent",
-  "version": "0.26.11",
+  "version": "0.26.13",
   "description": "A minimal, public domain AI CLI agent compatible with OpenCode's JSON interface. Bun-only runtime.",
   "main": "src/index.js",
   "type": "module",
@@ -17,6 +17,7 @@
     "format": "prettier --write . ../scripts",
     "format:check": "prettier --check . ../scripts",
     "check:file-size": "node ../scripts/check-file-size.mjs",
+    "check:model-drift": "bun ../scripts/check-model-drift.mjs",
     "check": "npm run lint && npm run format:check && npm run check:file-size",
     "prepare": "cd .. && husky || true",
     "changeset": "changeset",
@@ -50,54 +51,54 @@
     "LICENSE"
   ],
   "dependencies": {
-    "@agentclientprotocol/sdk": "^1.5.1",
-    "@ai-sdk/amazon-bedrock": "^5.0.102",
-    "@ai-sdk/anthropic": "^4.0.69",
-    "@ai-sdk/azure": "^4.0.86",
-    "@ai-sdk/google": "^4.0.86",
-    "@ai-sdk/google-vertex": "^5.0.99",
-    "@ai-sdk/groq": "^4.0.53",
-    "@ai-sdk/mcp": "^2.0.63",
-    "@ai-sdk/mistral": "^4.0.55",
-    "@ai-sdk/openai": "^4.0.82",
-    "@ai-sdk/openai-compatible": "^3.0.60",
-    "@ai-sdk/provider": "^4.0.20",
-    "@ai-sdk/xai": "^5.0.13",
+    "@agentclientprotocol/sdk": "^1.8.0",
+    "@ai-sdk/amazon-bedrock": "^5.0.113",
+    "@ai-sdk/anthropic": "^4.0.78",
+    "@ai-sdk/azure": "^4.0.99",
+    "@ai-sdk/google": "^4.0.93",
+    "@ai-sdk/google-vertex": "^5.0.109",
+    "@ai-sdk/groq": "^4.0.59",
+    "@ai-sdk/mcp": "^2.0.73",
+    "@ai-sdk/mistral": "^4.0.62",
+    "@ai-sdk/openai": "^4.0.91",
+    "@ai-sdk/openai-compatible": "^3.0.67",
+    "@ai-sdk/provider": "^4.0.26",
+    "@ai-sdk/xai": "^5.0.20",
     "@clack/prompts": "^1.8.1",
     "@hono/standard-validator": "^0.4.0",
     "@hono/zod-validator": "^0.9.1",
-    "@modelcontextprotocol/sdk": "^1.31.0",
+    "@modelcontextprotocol/sdk": "^1.32.1",
     "@octokit/graphql": "^9.0.5",
     "@octokit/rest": "^22.0.1",
     "@openauthjs/openauth": "^0.4.3",
-    "@opencode-ai/sdk": "^1.18.33",
+    "@opencode-ai/sdk": "^1.19.0",
     "@openrouter/ai-sdk-provider": "^3.1.0",
-    "@opentui/core": "0.5.13",
-    "@opentui/solid": "0.5.13",
+    "@opentui/core": "0.6.1",
+    "@opentui/solid": "0.6.1",
     "@parcel/watcher": "^2.6.0",
     "@solid-primitives/event-bus": "^1.1.4",
     "@standard-community/standard-json": "^0.3.6",
     "@standard-community/standard-openapi": "^0.2.10",
-    "@zip.js/zip.js": "^2.18.2",
-    "ai": "^7.0.123",
+    "@zip.js/zip.js": "^2.23.0",
+    "ai": "^7.0.137",
     "chokidar": "^5.0.0",
     "clipboardy": "^5.3.2",
-    "command-stream": "^1.3.0",
+    "command-stream": "^2.3.0",
     "decimal.js": "^10.6.0",
     "diff": "^9.0.0",
     "fuzzysort": "^4.0.2",
     "glob": "^13.0.6",
     "gpt-tokenizer": "^4.0.0",
     "gray-matter": "^4.0.3",
-    "hono": "^4.13.12",
-    "hono-openapi": "^1.3.3",
-    "ignore": "^7.0.10",
+    "hono": "^4.13.13",
+    "hono-openapi": "^1.3.5",
+    "ignore": "^7.0.12",
     "jsonc-parser": "^3.3.1",
     "lino-arguments": "^0.3.0",
-    "lino-objects-codec": "^0.8.0",
+    "lino-objects-codec": "^0.9.2",
     "log-lazy": "^1.0.4",
     "minimatch": "^10.2.6",
-    "remeda": "^2.50.0",
+    "remeda": "^2.51.0",
     "solid-js": "1.9.12",
     "strip-ansi": "^7.2.0",
     "test-anywhere": "^0.9.1",
@@ -109,18 +110,18 @@
     "web-tree-sitter": "0.25.10",
     "xdg-basedir": "^5.1.0",
     "yargs": "^18.2.0",
-    "zod": "^4.6.5"
+    "zod": "^4.6.6"
   },
   "devDependencies": {
     "@changesets/cli": "^3.0.3",
     "@eslint/js": "^10.0.1",
-    "eslint": "^10.11.0",
+    "eslint": "^10.12.0",
     "eslint-config-prettier": "^10.1.8",
     "eslint-plugin-prettier": "^5.5.6",
     "eslint-plugin-promise": "^7.3.0",
     "husky": "^9.1.7",
     "lint-staged": "^17.6.0",
-    "prettier": "^3.9.9"
+    "prettier": "^3.9.10"
   },
   "lint-staged": {
     "*.{js,mjs,cjs}": [
@@ -140,10 +141,11 @@
     "node-pty@1.2.0-beta.15": true
   },
   "overrides": {
+    "seroval": "^1.6.9",
     "undici": "^7.16.0"
   },
   "dependencyPins": {
     "solid-js": "https://github.com/anomalyco/opentui/issues/689",
-    "web-tree-sitter": "https://github.com/link-assistant/agent/issues/322"
+    "web-tree-sitter": "https://github.com/anomalyco/opentui/issues/1550"
   }
 }

package/src/cli/argv.ts

@@ -105,7 +105,7 @@ export function getCompactionSafetyMarginFromProcessArgv(): string | null {
 /**
  * Extract --compaction-models argument directly from process.argv
  * The value is a links notation references sequence, e.g.:
- *   "(big-pickle minimax-m2.5-free nemotron-3-super-free hy3-preview-free ling-2.6-flash-free gpt-5-nano same)"
+ *   "(kilo/nemotron-3-super-free kilo/nemotron-3-ultra-free kilo/nemotron-3.5-lightning-free same)"
  * @returns The compaction models argument from CLI or null if not found
  * @see https://github.com/link-assistant/agent/issues/232
  */

package/src/cli/run-options.js

@@ -190,7 +190,7 @@ export function buildRunOptions(yargs, defaultOptions = {}) {
     .option('compaction-model', {
       type: 'string',
       description:
-        'Model to use for context compaction and session summaries, in format providerID/modelID. Use "same" to use the base model. Default: opencode/gpt-5-nano (free, 400K context), which falls back to the --model model when --model points at another provider. Overridden by --compaction-models if both are specified.',
+        'Model to use for context compaction and session summaries, in format providerID/modelID. Use "same" to use the base model. Default: kilo/nemotron-3-super-free (free, Kilo authentication recommended, 262K context), which falls back to the --model model when --model points at another provider. Overridden by --compaction-models if both are specified.',
       default: defaultCompactionModel,
     })
     .option('compaction-models', {

package/src/config/defaults.ts

@@ -16,45 +16,28 @@ export interface DefaultConfigOptions {
 }
 
 /** Default model used when no `--model` CLI argument is provided. */
-export const DEFAULT_MODEL = 'opencode/minimax-m2.5-free';
+export const DEFAULT_MODEL = 'kilo/nemotron-3-super-free';
 
 /** Env var for overriding the default model in test runs and automation. */
 export const DEFAULT_MODEL_ENV = 'LINK_ASSISTANT_AGENT_DEFAULT_MODEL';
 
 /** Default compaction model used when no `--compaction-model` CLI argument is provided. */
-export const DEFAULT_COMPACTION_MODEL = 'opencode/gpt-5-nano';
+export const DEFAULT_COMPACTION_MODEL = 'kilo/nemotron-3-super-free';
 
 /** Env var for overriding the default compaction model in test runs and automation. */
 export const DEFAULT_COMPACTION_MODEL_ENV =
   'LINK_ASSISTANT_AGENT_DEFAULT_COMPACTION_MODEL';
 
 /**
- * Default compaction models cascade, ordered from smallest/cheapest context to largest.
- * During compaction, the system tries each model in order. If the used context exceeds
- * a model's context limit, it skips to the next larger model. If a model's rate limit
- * is reached, it also skips to the next model.
- * The special value "same" means use the same model as `--model`.
+ * Free Kilo models ordered by context capacity, followed by the session model.
+ * Compaction skips models whose context is too small or which hit a rate limit.
+ * Provider prefixes keep default secondary calls away from Zen's restricted tier.
+ * The special value "same" uses the model running the session.
  *
- * Parsed as links notation references sequence (single anonymous link):
- *   "(big-pickle minimax-m2.5-free nemotron-3-super-free hy3-preview-free ling-2.6-flash-free gpt-5-nano same)"
- *
- * Context limits (approximate):
- *   big-pickle:             ~200K
- *   minimax-m2.5-free:      ~204K
- *   nemotron-3-super-free:  ~204K
- *   hy3-preview-free:       ~256K
- *   ling-2.6-flash-free:    ~262K
- *   gpt-5-nano:             ~400K
- *   same:                   (base model's context)
- *
- * Note: qwen3.6-plus-free was removed because the free promotion ended in April 2026.
- * Note: minimax-m2.5-free is the default model again as of issue #266.
- * @see https://github.com/link-assistant/agent/issues/266
- * @see https://github.com/link-assistant/agent/issues/242
- * @see https://github.com/link-assistant/agent/issues/232
+ * @see https://github.com/link-assistant/agent/issues/327
  */
 export const DEFAULT_COMPACTION_MODELS =
-  '(big-pickle minimax-m2.5-free nemotron-3-super-free hy3-preview-free ling-2.6-flash-free gpt-5-nano same)';
+  '(kilo/nemotron-3-super-free kilo/nemotron-3-ultra-free kilo/nemotron-3.5-lightning-free same)';
 
 /** Env var for overriding the default compaction cascade in test runs and automation. */
 export const DEFAULT_COMPACTION_MODELS_ENV =

package/src/provider/free-tier-error.ts

@@ -0,0 +1,16 @@
+/** Zen uses this typed error for a permanent client restriction. */
+export function isFreeTierError(
+  status: number | undefined,
+  body: unknown
+): boolean {
+  if (status !== 403) return false;
+  try {
+    const parsed = typeof body === 'string' ? JSON.parse(body) : body;
+    return parsed?.error?.type === 'FreeTierError';
+  } catch {
+    return false;
+  }
+}
+
+export const FREE_TIER_ERROR_MESSAGE =
+  "OpenCode Zen's free tier is limited to OpenCode clients. Use kilo/nemotron-3-super-free (authenticate with agent auth login or KILO_API_KEY), or select a paid model.";

package/src/provider/kilo.ts

@@ -0,0 +1,49 @@
+import type { ModelsDev } from './models';
+
+/** Built-in aliases use exact IDs from https://api.kilo.ai/api/gateway/models. */
+export namespace Kilo {
+  export const MODELS_URL = 'https://api.kilo.ai/api/gateway/models';
+
+  function freeModel(
+    id: string,
+    name: string,
+    context: number,
+    output: number
+  ): ModelsDev.Model {
+    return {
+      id,
+      name,
+      release_date: '',
+      attachment: false,
+      reasoning: true,
+      temperature: true,
+      tool_call: true,
+      cost: { input: 0, output: 0, cache_read: 0, cache_write: 0 },
+      limit: { context, output },
+      modalities: { input: ['text'], output: ['text'] },
+      options: {},
+    };
+  }
+
+  // Super's output cap is deliberately conservative for existing integrations.
+  export const FREE_MODELS: Record<string, ModelsDev.Model> = {
+    'nemotron-3-super-free': freeModel(
+      'nvidia/nemotron-3-super-120b-a12b:free',
+      'Nemotron 3 Super (Free)',
+      262144,
+      32768
+    ),
+    'nemotron-3-ultra-free': freeModel(
+      'nvidia/nemotron-3-ultra-550b-a55b:free',
+      'Nemotron 3 Ultra (Free)',
+      1000000,
+      65536
+    ),
+    'nemotron-3.5-lightning-free': freeModel(
+      'nvidia/nemotron-3.5-lightning:free',
+      'Nemotron 3.5 Lightning (Free)',
+      1000000,
+      65536
+    ),
+  };
+}

package/src/provider/opencode-zen.ts

@@ -3,50 +3,29 @@ import type { ModelsDev } from './models';
 
 export namespace OpenCodeZen {
   const log = Log.create({ service: 'opencode-zen' });
-  const MODELS_URL = 'https://opencode.ai/zen/v1/models';
-  const DEPRECATED_FREE_MODEL_IDS = new Set(['trinity-large-preview-free']);
+  export const MODELS_URL = 'https://opencode.ai/zen/v1/models';
+  export const DEPRECATED_FREE_MODEL_IDS = new Set([
+    'trinity-large-preview-free',
+    'qwen3.6-plus-free',
+    'kimi-k2.5-free',
+    'minimax-m2.5-free',
+    'ling-2.6-flash-free',
+    'hy3-preview-free',
+    'nemotron-3-super-free',
+    'nemotron-3-ultra-free',
+  ]);
 
-  const FREE_MODEL_OVERRIDES: Record<
+  // Zen's free endpoints are restricted to OpenCode clients, so none is a default.
+  export const FREE_MODEL_OVERRIDES: Record<
     string,
     Partial<ModelsDev.Model> & { name: string }
   > = {
-    'minimax-m2.5-free': {
-      name: 'MiniMax M2.5 Free',
+    'nemotron-3.5-lightning-free': {
+      name: 'Nemotron 3.5 Lightning Free',
       reasoning: true,
-      release_date: '2026-02-12',
-      limit: { context: 204800, output: 131072 },
+      limit: { context: 1000000, output: 65536 },
       provider: { npm: '@ai-sdk/openai-compatible' },
     },
-    'ling-2.6-flash-free': {
-      name: 'Ling 2.6 Flash Free',
-      reasoning: false,
-      release_date: '2026-04-21',
-      limit: { context: 262100, output: 32800 },
-      provider: { npm: '@ai-sdk/openai-compatible' },
-    },
-    'hy3-preview-free': {
-      name: 'Hy3 Preview Free',
-      reasoning: true,
-      release_date: '2026-04-20',
-      limit: { context: 256000, output: 64000 },
-      provider: { npm: '@ai-sdk/openai-compatible' },
-    },
-    'nemotron-3-super-free': {
-      name: 'Nemotron 3 Super Free',
-      reasoning: true,
-      release_date: '2026-03-11',
-      limit: { context: 204800, output: 128000 },
-      provider: { npm: '@ai-sdk/openai-compatible' },
-    },
-    'gpt-5-nano': {
-      name: 'GPT-5 Nano',
-      attachment: true,
-      reasoning: true,
-      temperature: false,
-      release_date: '2025-08-07',
-      limit: { context: 400000, output: 128000 },
-      provider: { npm: '@ai-sdk/openai' },
-    },
     'big-pickle': {
       name: 'Big Pickle',
       reasoning: true,
@@ -68,11 +47,7 @@ export namespace OpenCodeZen {
 
   function isFreeModelID(modelID: string) {
     if (DEPRECATED_FREE_MODEL_IDS.has(modelID)) return false;
-    return (
-      modelID.endsWith('-free') ||
-      modelID === 'gpt-5-nano' ||
-      modelID === 'big-pickle'
-    );
+    return modelID.endsWith('-free') || modelID === 'big-pickle';
   }
 
   export async function fetchModelIDs(fetchFn: typeof fetch = fetch) {
@@ -115,6 +90,10 @@ export namespace OpenCodeZen {
     });
     if (!ids.has(modelID)) return undefined;
 
+    return freeModelInfo(modelID);
+  }
+
+  export function freeModelInfo(modelID: string): ModelsDev.Model {
     const overrides = FREE_MODEL_OVERRIDES[modelID];
     return {
       id: modelID,

package/src/provider/provider.ts

@@ -20,6 +20,7 @@ import { createCacheModel } from './cache';
 import { RetryFetch } from './retry-fetch';
 import { SSEUsageExtractor } from '../util/sse-usage-extractor';
 import { OpenCodeZen } from './opencode-zen';
+import { Kilo } from './kilo';
 
 // Direct imports for bundled providers - these are pre-installed to avoid runtime installation hangs
 // @see https://github.com/link-assistant/agent/issues/173
@@ -445,18 +446,13 @@ export namespace Provider {
       };
     },
     /**
-     * Kilo provider - access to 500+ AI models through Kilo Gateway
+     * Kilo provider - access to AI models through Kilo Gateway
      * Uses OpenRouter-compatible API at https://api.kilo.ai/api/openrouter
      *
-     * Authentication required: run `agent auth login` and select "Kilo Gateway"
+     * Authentication recommended: run `agent auth login` and select "Kilo Gateway"
      * For API key authentication, set KILO_API_KEY environment variable
      *
-     * Free models available after authentication:
-     * - GLM-5 (z-ai/glm-5) - Free limited time, flagship Z.AI model
-     * - GLM 4.7 (z-ai/glm-4.7:free) - Free, agent-centric model
-     * - Kimi K2.5 (moonshot/kimi-k2.5:free) - Free, agentic capabilities
-     * - MiniMax M2.1 (minimax/m2.1:free) - Free, general-purpose
-     * - Giga Potato (giga-potato:free) - Free evaluation model
+     * Free Nemotron models are listed in Kilo.FREE_MODELS.
      *
      * @see https://kilo.ai/docs/gateway
      * @see https://kilo.ai/docs/advanced-usage/free-and-budget-models
@@ -932,174 +928,24 @@ export namespace Provider {
       models: {}, // Models are dynamically created based on the provider/model syntax
     };
 
-    // Add Kilo provider for access to 500+ AI models through Kilo Gateway
-    // Free models available: GLM-5, GLM 4.7, Kimi K2.5, MiniMax M2.1, Giga Potato
-    // @see https://kilo.ai/docs/gateway
-    // @see https://github.com/link-assistant/agent/issues/159
-    database['kilo'] = {
+    // Remove withdrawn Zen entries even when models.dev has a stale cache.
+    // Operator model overrides are applied below and remain authoritative.
+    if (database.opencode) {
+      for (const id of OpenCodeZen.DEPRECATED_FREE_MODEL_IDS) {
+        delete database.opencode.models[id];
+      }
+      for (const id of Object.keys(OpenCodeZen.FREE_MODEL_OVERRIDES)) {
+        database.opencode.models[id] = OpenCodeZen.freeModelInfo(id);
+      }
+    }
+
+    database.kilo = {
       id: 'kilo',
       name: 'Kilo Gateway',
       npm: '@openrouter/ai-sdk-provider',
       api: 'https://api.kilo.ai/api/openrouter',
       env: ['KILO_API_KEY'],
-      models: {
-        // GLM-5 - Flagship Z.AI model, free tier
-        // Kilo API model ID: z-ai/glm-5:free (NOT z-ai/glm-5 which is paid)
-        'glm-5-free': {
-          id: 'z-ai/glm-5:free',
-          name: 'GLM-5 (Free)',
-          release_date: '2026-02-11',
-          attachment: false,
-          reasoning: true,
-          temperature: true,
-          tool_call: true,
-          cost: {
-            input: 0,
-            output: 0,
-            cache_read: 0,
-            cache_write: 0,
-          },
-          limit: {
-            context: 202800,
-            output: 131072,
-          },
-          modalities: {
-            input: ['text'],
-            output: ['text'],
-          },
-          options: {},
-        },
-        // GLM 4.5 Air - Free Z.AI model (replaces non-existent glm-4.7:free)
-        // Kilo API model ID: z-ai/glm-4.5-air:free
-        'glm-4.5-air-free': {
-          id: 'z-ai/glm-4.5-air:free',
-          name: 'GLM 4.5 Air (Free)',
-          release_date: '2026-01-15',
-          attachment: false,
-          reasoning: true,
-          temperature: true,
-          tool_call: true,
-          cost: {
-            input: 0,
-            output: 0,
-            cache_read: 0,
-            cache_write: 0,
-          },
-          limit: {
-            context: 131072,
-            output: 96000,
-          },
-          modalities: {
-            input: ['text'],
-            output: ['text'],
-          },
-          options: {},
-        },
-        // MiniMax M2.5 - General-purpose, free (upgraded from M2.1)
-        // Kilo API model ID: minimax/minimax-m2.5:free
-        'minimax-m2.5-free': {
-          id: 'minimax/minimax-m2.5:free',
-          name: 'MiniMax M2.5 (Free)',
-          release_date: '2026-01-01',
-          attachment: false,
-          reasoning: true,
-          temperature: true,
-          tool_call: true,
-          cost: {
-            input: 0,
-            output: 0,
-            cache_read: 0,
-            cache_write: 0,
-          },
-          limit: {
-            context: 204800,
-            output: 131072,
-          },
-          modalities: {
-            input: ['text'],
-            output: ['text'],
-          },
-          options: {},
-        },
-        // Giga Potato - Free evaluation model
-        // Kilo API model ID: giga-potato (no :free suffix)
-        'giga-potato-free': {
-          id: 'giga-potato',
-          name: 'Giga Potato (Free)',
-          release_date: '2026-01-01',
-          attachment: true,
-          reasoning: true,
-          temperature: true,
-          tool_call: true,
-          cost: {
-            input: 0,
-            output: 0,
-            cache_read: 0,
-            cache_write: 0,
-          },
-          limit: {
-            context: 256000,
-            output: 32000,
-          },
-          modalities: {
-            input: ['text', 'image'],
-            output: ['text'],
-          },
-          options: {},
-        },
-        // Trinity Large Preview - Preview model from Arcee AI
-        // Kilo API model ID: arcee-ai/trinity-large-preview:free
-        'trinity-large-preview': {
-          id: 'arcee-ai/trinity-large-preview:free',
-          name: 'Trinity Large Preview (Free)',
-          release_date: '2026-01-01',
-          attachment: false,
-          reasoning: false,
-          temperature: true,
-          tool_call: true,
-          cost: {
-            input: 0,
-            output: 0,
-            cache_read: 0,
-            cache_write: 0,
-          },
-          limit: {
-            context: 131000,
-            output: 65536,
-          },
-          modalities: {
-            input: ['text'],
-            output: ['text'],
-          },
-          options: {},
-        },
-        // DeepSeek R1 - Reasoning model, free
-        // Kilo API model ID: deepseek/deepseek-r1-0528:free
-        'deepseek-r1-free': {
-          id: 'deepseek/deepseek-r1-0528:free',
-          name: 'DeepSeek R1 (Free)',
-          release_date: '2025-05-28',
-          attachment: false,
-          reasoning: true,
-          temperature: true,
-          tool_call: false,
-          cost: {
-            input: 0,
-            output: 0,
-            cache_read: 0,
-            cache_write: 0,
-          },
-          limit: {
-            context: 163840,
-            output: 163840,
-          },
-          modalities: {
-            input: ['text'],
-            output: ['text'],
-          },
-          options: {},
-        },
-      },
+      models: structuredClone(Kilo.FREE_MODELS),
     };
 
     for (const [providerID, provider] of configProviders) {
@@ -1794,7 +1640,13 @@ export namespace Provider {
         await ModelsDev.refresh();
         const freshDB = await ModelsDev.get();
         const freshProvider = freshDB[providerID];
-        if (freshProvider?.models[modelID]) {
+        if (
+          freshProvider?.models[modelID] &&
+          !(
+            providerID === 'opencode' &&
+            OpenCodeZen.DEPRECATED_FREE_MODEL_IDS.has(modelID)
+          )
+        ) {
           // Model found after refresh - update provider info and use it
           provider.info.models[modelID] = freshProvider.models[modelID];
           info = freshProvider.models[modelID];
@@ -1958,23 +1810,10 @@ export namespace Provider {
       priority = priority.filter((m) => m !== 'claude-haiku-4.5');
     }
     if (providerID === 'opencode' || providerID === 'local') {
-      priority = [
-        'minimax-m2.5-free',
-        'ling-2.6-flash-free',
-        'hy3-preview-free',
-        'nemotron-3-super-free',
-        'gpt-5-nano',
-        'big-pickle',
-      ];
+      priority = Object.keys(OpenCodeZen.FREE_MODEL_OVERRIDES);
     }
     if (providerID === 'kilo') {
-      priority = [
-        'glm-5-free',
-        'glm-4.5-air-free',
-        'minimax-m2.5-free',
-        'deepseek-r1-free',
-        'giga-potato-free',
-      ];
+      priority = Object.keys(Kilo.FREE_MODELS);
     }
     for (const item of priority) {
       for (const model of Object.keys(provider.info.models)) {
@@ -1994,11 +1833,9 @@ export namespace Provider {
   const priority = [
     'big-pickle',
     'gpt-5-nano',
+    'nemotron-3.5-lightning-free',
+    'nemotron-3-ultra-free',
     'nemotron-3-super-free',
-    'hy3-preview-free',
-    'ling-2.6-flash-free',
-    'minimax-m2.5-free',
-    'glm-5-free',
     'gpt-5',
     'claude-sonnet-4',
     'gemini-3-pro',
@@ -2083,12 +1920,9 @@ export namespace Provider {
    * Resolve a short model name (without provider prefix) to the appropriate provider.
    * This function finds which provider should handle a model when no explicit provider is specified.
    *
-   * Priority for free models:
-   * 1. If model is uniquely available in one provider, use that provider
-   * 2. If model is available in multiple providers, prioritize based on free model availability:
-   *    - kilo: glm-5-free, glm-4.5-air-free, giga-potato-free, deepseek-r1-free (unique to Kilo)
-   *    - opencode: minimax-m2.5-free, ling-2.6-flash-free, hy3-preview-free, big-pickle, gpt-5-nano, nemotron-3-super-free
-   * 3. For shared models, prefer OpenCode first, then fall back to Kilo on rate limit
+   * Built-in Kilo free aliases prefer Kilo, including aliases also served by
+   * Zen, whose free tier is restricted to OpenCode clients. Other short names
+   * resolve against available provider catalogs.
    *
    * @param modelID - Short model name without provider prefix
    * @returns Provider ID that should handle this model, or undefined if not found
@@ -2098,25 +1932,9 @@ export namespace Provider {
   ): Promise<{ providerID: string; modelID: string } | undefined> {
     const s = await state();
 
-    // Define model-to-provider mappings for free models
-    // Models unique to Kilo (GLM models from Z.AI are only free on Kilo)
-    const kiloUniqueModels = [
-      'glm-5-free',
-      'glm-4.5-air-free',
-      'giga-potato-free',
-      'trinity-large-preview',
-      'deepseek-r1-free',
-    ];
-
-    // Check if it's a Kilo-unique model
-    if (kiloUniqueModels.includes(modelID)) {
-      const kiloProvider = s.providers['kilo'];
-      if (kiloProvider && kiloProvider.info.models[modelID]) {
-        registryLog.info(() => ({
-          message: 'resolved short model name to kilo (unique)',
-          name: modelID,
-          candidate: `kilo/${modelID}`,
-        }));
+    if (Object.hasOwn(Kilo.FREE_MODELS, modelID)) {
+      const kiloProvider = s.providers.kilo;
+      if (kiloProvider?.info.models[modelID]) {
         return { providerID: 'kilo', modelID };
       }
     }
@@ -2187,8 +2005,8 @@ export namespace Provider {
    * or a clear error explaining why it's not available.
    *
    * Examples:
-   * - "kilo/glm-5-free" -> { providerID: "kilo", modelID: "glm-5-free" }
-   * - "glm-5-free" -> { providerID: "kilo", modelID: "glm-5-free" } (resolved)
+   * - "kilo/nemotron-3-super-free" -> { providerID: "kilo", modelID: "nemotron-3-super-free" }
+   * - "nemotron-3-super-free" -> { providerID: "kilo", modelID: "nemotron-3-super-free" } (resolved)
    * - "big-pickle" -> { providerID: "opencode", modelID: "big-pickle" } (resolved)
    * - "nonexistent-model" -> throws ModelNotFoundError
    *
@@ -2249,10 +2067,10 @@ export namespace Provider {
    * When one provider hits rate limits, the system can try an alternative.
    *
    * Note: This is only used for models without explicit provider specification.
-   * If user specifies "kilo/deepseek-r1-free", no fallback will occur.
+   * If user specifies "kilo/nemotron-3-super-free", no fallback will occur.
    */
   const SHARED_FREE_MODELS: Record<string, string[]> = {
-    'minimax-m2.5-free': ['opencode', 'kilo'],
+    'nemotron-3.5-lightning-free': ['kilo', 'opencode'],
   };
 
   /**
@@ -2260,7 +2078,7 @@ export namespace Provider {
    * This function returns a list of alternative providers that offer the same model.
    *
    * Note: This only returns alternatives for models without explicit provider specification.
-   * If the original request had an explicit provider (like "kilo/deepseek-r1-free"), this returns empty array.
+   * If the original request had an explicit provider (like "kilo/nemotron-3-super-free"), this returns empty array.
    *
    * @param modelID - The model ID to find alternatives for
    * @param failedProviderID - The provider that failed

package/src/session/message-v2.ts

@@ -1,3 +1,7 @@
+import {
+  isFreeTierError,
+  FREE_TIER_ERROR_MESSAGE,
+} from '../provider/free-tier-error';
 import z from 'zod';
 import { Bus } from '../bus';
 import { NamedError } from '../util/error';
@@ -866,7 +870,12 @@ export namespace MessageV2 {
         ? (root['error'] as Record<string, unknown>)
         : root;
     const rawCode =
-      candidate['code'] ?? candidate['statusCode'] ?? candidate['status'];
+      candidate['code'] ??
+      candidate['statusCode'] ??
+      candidate['status'] ??
+      root['code'] ??
+      root['statusCode'] ??
+      root['status'];
     const statusCode =
       typeof rawCode === 'number'
         ? rawCode
@@ -900,6 +909,27 @@ export namespace MessageV2 {
   }
 
   export function fromError(e: unknown, ctx: { providerID: string }) {
+    const providerError = APICallError.isInstance(e)
+      ? undefined
+      : normalizeProviderErrorObject(e);
+    const freeTierRefusal =
+      ctx.providerID === 'opencode' &&
+      (APICallError.isInstance(e)
+        ? isFreeTierError(e.statusCode, e.responseBody)
+        : isFreeTierError(providerError?.statusCode, e));
+    if (freeTierRefusal) {
+      return new MessageV2.APIError(
+        {
+          message: FREE_TIER_ERROR_MESSAGE,
+          statusCode: 403,
+          isRetryable: false,
+          responseBody: APICallError.isInstance(e)
+            ? e.responseBody
+            : providerError?.responseBody,
+        },
+        { cause: e }
+      ).toObject();
+    }
     switch (true) {
       case e instanceof DOMException && e.name === 'AbortError':
         return new MessageV2.AbortedError(