refactor: make AI runtime z.ai-only and default to glm-5

.env.example

@@ -16,15 +16,9 @@ BETTER_AUTH_TRUSTED_ORIGINS=https://fiscal.b11studio.xyz
 # Legacy OPENCLAW_* variables are removed and no longer read by the app.
 # Coding endpoint is hardcoded in runtime: https://api.z.ai/api/coding/paas/v4
 ZHIPU_API_KEY=
-ZHIPU_MODEL=glm-4.7-flashx
+ZHIPU_MODEL=glm-5
 AI_TEMPERATURE=0.2
 
-# Local extraction model (Ollama, OpenAI-compatible API)
-# For host Ollama from Docker, use http://host.docker.internal:11434
-OLLAMA_BASE_URL=http://127.0.0.1:11434
-OLLAMA_MODEL=qwen3:8b
-OLLAMA_API_KEY=ollama
-
 # SEC API etiquette
 SEC_USER_AGENT=Fiscal Clone <support@fiscal.local>
 

README.md

@@ -14,9 +14,7 @@ Turbopack-first rebuild of a fiscal.ai-style terminal with Vercel AI SDK integra
 - Eden Treaty for type-safe frontend API calls
 - Workflow DevKit Postgres World for background task execution durability
 - SQLite-backed app domain storage (watchlist, holdings, filings, task projection, insights)
-- Vercel AI SDK (`ai`) with dual-model routing:
+- Vercel AI SDK (`ai`) with Zhipu (`zhipu-ai-provider`) via Coding API (`https://api.z.ai/api/coding/paas/v4`)
-  - Ollama (`@ai-sdk/openai`) for lightweight filing extraction/parsing
-  - Zhipu (`zhipu-ai-provider`) for heavyweight narrative reports (`https://api.z.ai/api/coding/paas/v4`)
 
 ## Run locally
 
@@ -47,8 +45,7 @@ docker compose up --build -d
 ```
 
 For local Docker, host port mapping comes from `docker-compose.override.yml` (default `http://localhost:3000` via `APP_PORT`).
-The app calls Zhipu directly via AI SDK for heavy reports and calls Ollama for lightweight filing extraction.
+The app calls Zhipu directly via AI SDK for extraction and report generation.
-When running in Docker and Ollama runs on the host, set `OLLAMA_BASE_URL=http://host.docker.internal:11434`.
 Zhipu always targets the Coding API endpoint (`https://api.z.ai/api/coding/paas/v4`).
 On container startup, the app applies Drizzle migrations automatically before launching Next.js.
 The app stores SQLite data in Docker volume `fiscal_sqlite_data` (mounted to `/app/data`) and workflow world data in Postgres volume `workflow_postgres_data`.
@@ -100,13 +97,10 @@ BETTER_AUTH_BASE_URL=https://fiscal.b11studio.xyz
 BETTER_AUTH_TRUSTED_ORIGINS=https://fiscal.b11studio.xyz
 
 ZHIPU_API_KEY=
-ZHIPU_MODEL=glm-4.7-flashx
+ZHIPU_MODEL=glm-5
 # optional generation tuning
 AI_TEMPERATURE=0.2
 
-OLLAMA_BASE_URL=http://127.0.0.1:11434
-OLLAMA_MODEL=qwen3:8b
-OLLAMA_API_KEY=ollama
 SEC_USER_AGENT=Fiscal Clone <support@fiscal.local>
 
 WORKFLOW_TARGET_WORLD=@workflow/world-postgres
@@ -119,8 +113,7 @@ WORKFLOW_LOCAL_DATA_DIR=.workflow-data
 WORKFLOW_LOCAL_QUEUE_CONCURRENCY=100
 ```
 
-If `ZHIPU_API_KEY` is unset, the app uses local fallback analysis so task workflows still run.
+`ZHIPU_API_KEY` is required for AI workloads (extraction and report generation). Missing or invalid credentials fail AI tasks.
-If Ollama is unavailable, filing extraction falls back to deterministic metadata-based extraction and still proceeds to heavy report generation.
 `ZHIPU_BASE_URL` is deprecated and ignored; runtime always uses `https://api.z.ai/api/coding/paas/v4`.
 
 ## API surface

bun.lock

@@ -5,7 +5,6 @@
     "": {
       "name": "fiscal-frontend",
       "dependencies": {
-        "@ai-sdk/openai": "^2.0.62",
         "@elysiajs/eden": "^1.4.8",
         "@libsql/client": "^0.17.0",
         "@tailwindcss/postcss": "^4.2.1",
@@ -42,11 +41,9 @@
   "packages": {
     "@ai-sdk/gateway": ["@ai-sdk/gateway@3.0.58", "", { "dependencies": { "@ai-sdk/provider": "3.0.8", "@ai-sdk/provider-utils": "4.0.15", "@vercel/oidc": "3.1.0" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-2e1hBCKsd+7m0hELwrakR1QDfZfFhz9PF2d4qb8TxQueEyApo7ydlEWRpXeKC+KdA2FRV21dMb1G6FxdeNDa2w=="],
 
-    "@ai-sdk/openai": ["@ai-sdk/openai@2.0.95", "", { "dependencies": { "@ai-sdk/provider": "2.0.1", "@ai-sdk/provider-utils": "3.0.21" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-2CABPaa1UNh7dPyZUIB/Dc4AbvJioFnmryRx45sx7ezBSOdR0zxG6gbrSd/fZ0GVbptSZeLmF9omu10d/GxmJA=="],
+    "@ai-sdk/provider": ["@ai-sdk/provider@3.0.8", "", { "dependencies": { "json-schema": "^0.4.0" } }, "sha512-oGMAgGoQdBXbZqNG0Ze56CHjDZ1IDYOwGYxYjO5KLSlz5HiNQ9udIXsPZ61VWaHGZ5XW/jyjmr6t2xz2jGVwbQ=="],
 
-    "@ai-sdk/provider": ["@ai-sdk/provider@2.0.1", "", { "dependencies": { "json-schema": "^0.4.0" } }, "sha512-KCUwswvsC5VsW2PWFqF8eJgSCu5Ysj7m1TxiHTVA6g7k360bk0RNQENT8KTMAYEs+8fWPD3Uu4dEmzGHc+jGng=="],
+    "@ai-sdk/provider-utils": ["@ai-sdk/provider-utils@4.0.15", "", { "dependencies": { "@ai-sdk/provider": "3.0.8", "@standard-schema/spec": "^1.1.0", "eventsource-parser": "^3.0.6" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-8XiKWbemmCbvNN0CLR9u3PQiet4gtEVIrX4zzLxnCj06AwsEDJwJVBbKrEI4t6qE8XRSIvU2irka0dcpziKW6w=="],
-
-    "@ai-sdk/provider-utils": ["@ai-sdk/provider-utils@3.0.21", "", { "dependencies": { "@ai-sdk/provider": "2.0.1", "@standard-schema/spec": "^1.0.0", "eventsource-parser": "^3.0.6" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-veuMwTLxsgh31Jjn0SnBABnM1f7ebHhRWcV2ZuY3hP3iJDCZ8VXBaYqcHXoOQDqUXTCas08sKQcHyWK+zl882Q=="],
 
     "@alloc/quick-lru": ["@alloc/quick-lru@5.2.0", "", {}, "sha512-UrcABB+4bUrFABwbluTIBErXwvbsU/V7TZWfmbgJfbkwiBuziS9gxdODUyuiecfdGQ85jglMW6juS3+z5TsKLw=="],
 
@@ -1528,10 +1525,6 @@
 
     "zod": ["zod@4.3.6", "", {}, "sha512-rftlrkhHZOcjDwkGlnUtZZkvaPHCsDATp4pGpuOOMDaTdDDXF91wuVDJoWoPsKX/3YPQ5fHuF3STjcYyKr+Qhg=="],
 
-    "@ai-sdk/gateway/@ai-sdk/provider": ["@ai-sdk/provider@3.0.8", "", { "dependencies": { "json-schema": "^0.4.0" } }, "sha512-oGMAgGoQdBXbZqNG0Ze56CHjDZ1IDYOwGYxYjO5KLSlz5HiNQ9udIXsPZ61VWaHGZ5XW/jyjmr6t2xz2jGVwbQ=="],
-
-    "@ai-sdk/gateway/@ai-sdk/provider-utils": ["@ai-sdk/provider-utils@4.0.15", "", { "dependencies": { "@ai-sdk/provider": "3.0.8", "@standard-schema/spec": "^1.1.0", "eventsource-parser": "^3.0.6" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-8XiKWbemmCbvNN0CLR9u3PQiet4gtEVIrX4zzLxnCj06AwsEDJwJVBbKrEI4t6qE8XRSIvU2irka0dcpziKW6w=="],
-
     "@ai-sdk/gateway/@vercel/oidc": ["@vercel/oidc@3.1.0", "", {}, "sha512-Fw28YZpRnA3cAHHDlkt7xQHiJ0fcL+NRcIqsocZQUSmbzeIKRpwttJjik5ZGanXP+vlA4SbTg+AbA3bP363l+w=="],
 
     "@aws-crypto/sha256-browser/@aws-sdk/types": ["@aws-sdk/types@3.973.3", "", { "dependencies": { "@smithy/types": "^4.13.0", "tslib": "^2.6.2" } }, "sha512-tma6D8/xHZHJEUqmr6ksZjZ0onyIUqKDQLyp50ttZJmS0IwFYzxBgp5CxFvpYAnah52V3UtgrqGA6E83gtT7NQ=="],
@@ -1786,10 +1779,6 @@
 
     "@xhmikosr/downloader/file-type": ["file-type@20.5.0", "", { "dependencies": { "@tokenizer/inflate": "^0.2.6", "strtok3": "^10.2.0", "token-types": "^6.0.0", "uint8array-extras": "^1.4.0" } }, "sha512-BfHZtG/l9iMm4Ecianu7P8HRD2tBHLtjXinm4X62XBOYzi7CYA7jyqfJzOvXHqzVrVPYqBo2/GvbARMaaJkKVg=="],
 
-    "ai/@ai-sdk/provider": ["@ai-sdk/provider@3.0.8", "", { "dependencies": { "json-schema": "^0.4.0" } }, "sha512-oGMAgGoQdBXbZqNG0Ze56CHjDZ1IDYOwGYxYjO5KLSlz5HiNQ9udIXsPZ61VWaHGZ5XW/jyjmr6t2xz2jGVwbQ=="],
-
-    "ai/@ai-sdk/provider-utils": ["@ai-sdk/provider-utils@4.0.15", "", { "dependencies": { "@ai-sdk/provider": "3.0.8", "@standard-schema/spec": "^1.1.0", "eventsource-parser": "^3.0.6" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-8XiKWbemmCbvNN0CLR9u3PQiet4gtEVIrX4zzLxnCj06AwsEDJwJVBbKrEI4t6qE8XRSIvU2irka0dcpziKW6w=="],
-
     "ansi-escapes/type-fest": ["type-fest@0.21.3", "", {}, "sha512-t0rzBq87m3fVcduHDUFhKmyyX+9eo6WQjZvf51Ea/M0Q7+T374Jp1aUiyUl0GKxp8M/OETVHSDvmkyPgvX+X2w=="],
 
     "body-parser/debug": ["debug@2.6.9", "", { "dependencies": { "ms": "2.0.0" } }, "sha512-bC7ElrdJaJnPbAP+1EotYvqZsb3ecl5wi6Bfi6BJTUcNowp6cvspg0jXznRTKDjm/E7AdgFBVeAPVMNcKGsHMA=="],
@@ -1872,6 +1861,10 @@
 
     "wsl-utils/is-wsl": ["is-wsl@3.1.1", "", { "dependencies": { "is-inside-container": "^1.0.0" } }, "sha512-e6rvdUCiQCAuumZslxRJWR/Doq4VpPR82kqclvcS0efgt430SlGIk05vdCN58+VrzgtIcfNODjozVielycD4Sw=="],
 
+    "zhipu-ai-provider/@ai-sdk/provider": ["@ai-sdk/provider@2.0.1", "", { "dependencies": { "json-schema": "^0.4.0" } }, "sha512-KCUwswvsC5VsW2PWFqF8eJgSCu5Ysj7m1TxiHTVA6g7k360bk0RNQENT8KTMAYEs+8fWPD3Uu4dEmzGHc+jGng=="],
+
+    "zhipu-ai-provider/@ai-sdk/provider-utils": ["@ai-sdk/provider-utils@3.0.21", "", { "dependencies": { "@ai-sdk/provider": "2.0.1", "@standard-schema/spec": "^1.0.0", "eventsource-parser": "^3.0.6" }, "peerDependencies": { "zod": "^3.25.76 || ^4.1.8" } }, "sha512-veuMwTLxsgh31Jjn0SnBABnM1f7ebHhRWcV2ZuY3hP3iJDCZ8VXBaYqcHXoOQDqUXTCas08sKQcHyWK+zl882Q=="],
+
     "@aws-crypto/sha256-browser/@aws-sdk/types/@smithy/types": ["@smithy/types@4.13.0", "", { "dependencies": { "tslib": "^2.6.2" } }, "sha512-COuLsZILbbQsdrwKQpkkpyep7lCsByxwj7m0Mg5v66/ZTyenlfBc40/QFQ5chO0YN/PNEH1Bi3fGtfXPnYNeDw=="],
 
     "@aws-crypto/sha256-browser/@smithy/util-utf8/@smithy/util-buffer-from": ["@smithy/util-buffer-from@2.2.0", "", { "dependencies": { "@smithy/is-array-buffer": "^2.2.0", "tslib": "^2.6.2" } }, "sha512-IJdWBbTcMQ6DA0gdNhh/BwrLkDR+ADW5Kr1aZmd4k3DIF6ezMV4R2NIAmT08wQJ3yUK82thHWmC/TnK/wpMMIA=="],

docker-compose.yml

@@ -37,11 +37,8 @@ services:
       BETTER_AUTH_ADMIN_USER_IDS: ${BETTER_AUTH_ADMIN_USER_IDS:-}
       BETTER_AUTH_TRUSTED_ORIGINS: ${BETTER_AUTH_TRUSTED_ORIGINS:-https://fiscal.b11studio.xyz}
       ZHIPU_API_KEY: ${ZHIPU_API_KEY:-}
-      ZHIPU_MODEL: ${ZHIPU_MODEL:-glm-4.7-flashx}
+      ZHIPU_MODEL: ${ZHIPU_MODEL:-glm-5}
       AI_TEMPERATURE: ${AI_TEMPERATURE:-0.2}
-      OLLAMA_BASE_URL: ${OLLAMA_BASE_URL:-http://127.0.0.1:11434}
-      OLLAMA_MODEL: ${OLLAMA_MODEL:-qwen3:8b}
-      OLLAMA_API_KEY: ${OLLAMA_API_KEY:-ollama}
       SEC_USER_AGENT: ${SEC_USER_AGENT:-Fiscal Clone <support@fiscal.local>}
       WORKFLOW_TARGET_WORLD: ${WORKFLOW_TARGET_WORLD:-@workflow/world-postgres}
       WORKFLOW_POSTGRES_URL: ${WORKFLOW_POSTGRES_URL:-postgres://workflow:workflow@workflow-postgres:5432/workflow}

lib/server/ai.test.ts

@@ -21,9 +21,10 @@ describe('ai config and runtime', () => {
       warn: () => {}
     });
 
+    expect(config.provider).toBe('zhipu');
     expect(config.apiKey).toBe('key');
     expect(config.baseUrl).toBe(CODING_API_BASE_URL);
-    expect(config.model).toBe('glm-4.7-flashx');
+    expect(config.model).toBe('glm-5');
     expect(config.temperature).toBe(0.2);
   });
 
@@ -39,7 +40,7 @@ describe('ai config and runtime', () => {
     expect(config.baseUrl).toBe(CODING_API_BASE_URL);
   });
 
-  it('clamps temperature into [0, 2]', () => {
+  it('clamps report temperature into [0, 2]', () => {
     const negative = getAiConfig({
       env: {
         ZHIPU_API_KEY: 'key',
@@ -68,23 +69,50 @@ describe('ai config and runtime', () => {
     expect(invalid.temperature).toBe(0.2);
   });
 
-  it('returns fallback output when ZHIPU_API_KEY is missing', async () => {
+  it('uses extraction workload with zhipu config and zero temperature', async () => {
-    const generate = mock(async () => ({ text: 'should-not-be-used' }));
+    const createModel = mock((config: {
+      provider: string;
+      apiKey?: string;
+      model: string;
+      baseUrl: string;
+      temperature: number;
+    }) => {
+      expect(config.provider).toBe('zhipu');
+      expect(config.apiKey).toBe('new-key');
+      expect(config.baseUrl).toBe(CODING_API_BASE_URL);
+      expect(config.model).toBe('glm-5');
+      expect(config.temperature).toBe(0);
+      return { modelId: config.model };
+    });
+    const generate = mock(async (input: {
+      model: unknown;
+      system?: string;
+      prompt: string;
+      temperature: number;
+      maxRetries?: number;
+    }) => {
+      expect(input.system).toBe('Return strict JSON only.');
+      expect(input.prompt).toBe('Extract this filing');
+      expect(input.temperature).toBe(0);
+      expect(input.maxRetries).toBe(0);
+      return { text: '{"summary":"ok"}' };
+    });
 
-    const result = await runAiAnalysis(
+    const result = await runAiAnalysis('Extract this filing', 'Return strict JSON only.', {
-      'Prompt line one\nPrompt line two',
+      env: {
-      'System prompt',
+        ZHIPU_API_KEY: 'new-key'
-      {
+      },
-        env: {},
       warn: () => {},
+      workload: 'extraction',
+      createModel,
       generate
-      }
+    });
-    );
 
-    expect(result.provider).toBe('local-fallback');
+    expect(result.provider).toBe('zhipu');
-    expect(result.model).toBe('glm-4.7-flashx');
+    expect(result.model).toBe('glm-5');
-    expect(result.text).toContain('AI SDK fallback mode is active');
+    expect(result.text).toBe('{"summary":"ok"}');
-    expect(generate).not.toHaveBeenCalled();
+    expect(createModel).toHaveBeenCalledTimes(1);
+    expect(generate).toHaveBeenCalledTimes(1);
   });
 
   it('warns once when ZHIPU_BASE_URL is set because coding endpoint is hardcoded', () => {
@@ -103,11 +131,13 @@ describe('ai config and runtime', () => {
 
   it('uses configured ZHIPU values and injected generator when API key exists', async () => {
     const createModel = mock((config: {
+      provider: string;
       apiKey?: string;
       model: string;
       baseUrl: string;
       temperature: number;
     }) => {
+      expect(config.provider).toBe('zhipu');
       expect(config.apiKey).toBe('new-key');
       expect(config.baseUrl).toBe(CODING_API_BASE_URL);
       expect(config.model).toBe('glm-4-plus');
@@ -147,6 +177,29 @@ describe('ai config and runtime', () => {
     expect(result.text).toBe('Generated insight');
   });
 
+  it('throws when report workload runs without ZHIPU_API_KEY', async () => {
+    await expect(
+      runAiAnalysis('Analyze this filing', undefined, {
+        env: {},
+        warn: () => {},
+        createModel: () => ({}),
+        generate: async () => ({ text: 'should-not-be-used' })
+      })
+    ).rejects.toThrow('ZHIPU_API_KEY is required for AI workloads');
+  });
+
+  it('throws when extraction workload runs without ZHIPU_API_KEY', async () => {
+    await expect(
+      runAiAnalysis('Extract this filing', 'Return strict JSON only.', {
+        env: {},
+        warn: () => {},
+        workload: 'extraction',
+        createModel: () => ({}),
+        generate: async () => ({ text: 'should-not-be-used' })
+      })
+    ).rejects.toThrow('ZHIPU_API_KEY is required for AI workloads');
+  });
+
   it('throws when AI generation returns an empty response', async () => {
     await expect(
       runAiAnalysis('Analyze this filing', undefined, {
@@ -158,112 +211,7 @@ describe('ai config and runtime', () => {
     ).rejects.toThrow('AI SDK returned an empty response');
   });
 
-  it('uses ollama defaults for extraction workload config', () => {
+  it('keeps throwing unknown provider errors', async () => {
-    const config = getExtractionAiConfig({
-      env: {},
-      warn: () => {}
-    });
-
-    expect(config.provider).toBe('ollama');
-    expect(config.baseUrl).toBe('http://127.0.0.1:11434');
-    expect(config.model).toBe('qwen3:8b');
-    expect(config.apiKey).toBe('ollama');
-    expect(config.temperature).toBe(0);
-  });
-
-  it('uses extraction workload and returns ollama provider on success', async () => {
-    const createModel = mock((config: {
-      provider: string;
-      apiKey?: string;
-      model: string;
-      baseUrl: string;
-      temperature: number;
-    }) => {
-      expect(config.provider).toBe('ollama');
-      expect(config.baseUrl).toBe('http://127.0.0.1:11434');
-      expect(config.model).toBe('qwen3:8b');
-      expect(config.temperature).toBe(0);
-      return { modelId: config.model };
-    });
-    const generate = mock(async () => ({ text: '{"summary":"ok","keyPoints":[],"redFlags":[],"followUpQuestions":[],"portfolioSignals":[],"confidence":0.6}' }));
-
-    const result = await runAiAnalysis('Extract this filing', 'Return JSON', {
-      env: {
-        OLLAMA_MODEL: 'qwen3:8b'
-      },
-      warn: () => {},
-      workload: 'extraction',
-      createModel,
-      generate
-    });
-
-    expect(createModel).toHaveBeenCalledTimes(1);
-    expect(generate).toHaveBeenCalledTimes(1);
-    expect(result.provider).toBe('ollama');
-    expect(result.model).toBe('qwen3:8b');
-  });
-
-  it('falls back to local text when extraction workload generation fails', async () => {
-    const result = await runAiAnalysis('Extract this filing', 'Return JSON', {
-      env: {},
-      warn: () => {},
-      workload: 'extraction',
-      createModel: () => ({}),
-      generate: async () => {
-        throw new Error('ollama unavailable');
-      }
-    });
-
-    expect(result.provider).toBe('local-fallback');
-    expect(result.model).toBe('qwen3:8b');
-    expect(result.text).toContain('AI SDK fallback mode is active');
-  });
-
-  it('falls back to local text when report workload fails with insufficient balance', async () => {
-    const warn = mock((_message: string) => {});
-
-    const result = await runAiAnalysis('Analyze this filing', 'Use concise style', {
-      env: {
-        ZHIPU_API_KEY: 'new-key'
-      },
-      warn,
-      createModel: () => ({}),
-      generate: async () => {
-        throw new Error('AI_RetryError: Failed after 3 attempts. Last error: Insufficient balance or no resource package. Please recharge.');
-      }
-    });
-
-    expect(result.provider).toBe('local-fallback');
-    expect(result.model).toBe('glm-4.7-flashx');
-    expect(result.text).toContain('AI SDK fallback mode is active');
-    expect(warn).toHaveBeenCalledTimes(1);
-  });
-
-  it('falls back to local text when report workload cause contains insufficient balance', async () => {
-    const warn = mock((_message: string) => {});
-
-    const result = await runAiAnalysis('Analyze this filing', 'Use concise style', {
-      env: {
-        ZHIPU_API_KEY: 'new-key'
-      },
-      warn,
-      createModel: () => ({}),
-      generate: async () => {
-        const retryError = new Error('AI_RetryError: Failed after 3 attempts.');
-        (retryError as Error & { cause?: unknown }).cause = new Error(
-          'Last error: Insufficient balance or no resource package. Please recharge.'
-        );
-        throw retryError;
-      }
-    });
-
-    expect(result.provider).toBe('local-fallback');
-    expect(result.model).toBe('glm-4.7-flashx');
-    expect(result.text).toContain('AI SDK fallback mode is active');
-    expect(warn).toHaveBeenCalledTimes(1);
-  });
-
-  it('keeps throwing unknown report workload errors', async () => {
     await expect(
       runAiAnalysis('Analyze this filing', 'Use concise style', {
         env: {
@@ -277,4 +225,21 @@ describe('ai config and runtime', () => {
       })
     ).rejects.toThrow('unexpected schema mismatch');
   });
+
+  it('returns extraction config with same zhipu model and zero temperature', () => {
+    const config = getExtractionAiConfig({
+      env: {
+        ZHIPU_API_KEY: 'new-key',
+        ZHIPU_MODEL: 'glm-4-plus',
+        AI_TEMPERATURE: '0.9'
+      },
+      warn: () => {}
+    });
+
+    expect(config.provider).toBe('zhipu');
+    expect(config.apiKey).toBe('new-key');
+    expect(config.baseUrl).toBe(CODING_API_BASE_URL);
+    expect(config.model).toBe('glm-4-plus');
+    expect(config.temperature).toBe(0);
+  });
 });

lib/server/ai.ts

@@ -1,9 +1,8 @@
-import { createOpenAI } from '@ai-sdk/openai';
 import { generateText } from 'ai';
 import { createZhipu } from 'zhipu-ai-provider';
 
 type AiWorkload = 'report' | 'extraction';
-type AiProvider = 'zhipu' | 'ollama';
+type AiProvider = 'zhipu';
 
 type AiConfig = {
   provider: AiProvider;
@@ -39,9 +38,6 @@ type RunAiAnalysisOptions = GetAiConfigOptions & {
 };
 
 const CODING_API_BASE_URL = 'https://api.z.ai/api/coding/paas/v4';
-const OLLAMA_BASE_URL = 'http://127.0.0.1:11434';
-const OLLAMA_MODEL = 'qwen3:8b';
-const OLLAMA_API_KEY = 'ollama';
 
 let warnedIgnoredZhipuBaseUrl = false;
 
@@ -80,128 +76,13 @@ function warnIgnoredZhipuBaseUrl(env: EnvSource, warn: (message: string) => void
   );
 }
 
-function fallbackResponse(prompt: string) {
-  const clipped = prompt.split('\n').slice(0, 6).join(' ').slice(0, 260);
-
-  return [
-    'AI SDK fallback mode is active (live model configuration is missing or unavailable).',
-    'Thesis: Portfolio remains analyzable with local heuristics until live model access is configured.',
-    'Risk scan: Concentration and filing sentiment should be monitored after each sync cycle.',
-    `Context digest: ${clipped}`
-  ].join('\n\n');
-}
-
-function toOpenAiCompatibleBaseUrl(baseUrl: string) {
-  const normalized = baseUrl.endsWith('/')
-    ? baseUrl.slice(0, -1)
-    : baseUrl;
-
-  return normalized.endsWith('/v1')
-    ? normalized
-    : `${normalized}/v1`;
-}
-
-function asErrorMessage(error: unknown) {
-  if (error instanceof Error && error.message) {
-    return error.message;
-  }
-
-  return String(error);
-}
-
-function errorSearchText(error: unknown) {
-  const chunks: string[] = [];
-  const seen = new Set<unknown>();
-
-  const visit = (value: unknown) => {
-    if (value === null || value === undefined) {
-      return;
-    }
-
-    if (typeof value === 'string') {
-      const normalized = value.trim();
-      if (normalized.length > 0) {
-        chunks.push(normalized);
-      }
-
-      return;
-    }
-
-    if (typeof value !== 'object') {
-      chunks.push(String(value));
-      return;
-    }
-
-    if (seen.has(value)) {
-      return;
-    }
-    seen.add(value);
-
-    if (value instanceof Error) {
-      if (value.message) {
-        chunks.push(value.message);
-      }
-
-      const withCause = value as Error & { cause?: unknown };
-      if (withCause.cause !== undefined) {
-        visit(withCause.cause);
-      }
-      return;
-    }
-
-    const record = value as Record<string, unknown>;
-    visit(record.message);
-    visit(record.error);
-    visit(record.reason);
-    visit(record.detail);
-    visit(record.details);
-    visit(record.cause);
-  };
-
-  visit(error);
-  return chunks.join('\n');
-}
-
-const REPORT_FALLBACK_ERROR_PATTERNS: RegExp[] = [
-  /insufficient balance/i,
-  /no resource package/i,
-  /insufficient quota/i,
-  /quota exceeded/i,
-  /insufficient credit/i,
-  /invalid api key/i,
-  /authentication/i,
-  /unauthorized/i,
-  /forbidden/i,
-  /payment required/i,
-  /recharge/i,
-  /unable to connect/i,
-  /network/i,
-  /timeout/i,
-  /timed out/i,
-  /econnrefused/i
-];
-
-function shouldFallbackReportError(error: unknown) {
-  const searchText = errorSearchText(error) || asErrorMessage(error);
-  return REPORT_FALLBACK_ERROR_PATTERNS.some((pattern) => pattern.test(searchText));
-}
-
 function defaultCreateModel(config: AiConfig) {
-  if (config.provider === 'zhipu') {
   const zhipu = createZhipu({
     apiKey: config.apiKey,
     baseURL: config.baseUrl
   });
 
   return zhipu(config.model);
-  }
-
-  const openai = createOpenAI({
-    apiKey: config.apiKey ?? OLLAMA_API_KEY,
-    baseURL: toOpenAiCompatibleBaseUrl(config.baseUrl)
-  });
-
-  return openai.chat(config.model);
 }
 
 async function defaultGenerate(input: AiGenerateInput): Promise<AiGenerateOutput> {
@@ -228,21 +109,16 @@ export function getReportAiConfig(options?: GetAiConfigOptions) {
     provider: 'zhipu',
     apiKey: envValue('ZHIPU_API_KEY', env),
     baseUrl: CODING_API_BASE_URL,
-    model: envValue('ZHIPU_MODEL', env) ?? 'glm-4.7-flashx',
+    model: envValue('ZHIPU_MODEL', env) ?? 'glm-5',
     temperature: parseTemperature(envValue('AI_TEMPERATURE', env))
   } satisfies AiConfig;
 }
 
 export function getExtractionAiConfig(options?: GetAiConfigOptions) {
-  const env = options?.env ?? process.env;
-
   return {
-    provider: 'ollama',
+    ...getReportAiConfig(options),
-    apiKey: envValue('OLLAMA_API_KEY', env) ?? OLLAMA_API_KEY,
-    baseUrl: envValue('OLLAMA_BASE_URL', env) ?? OLLAMA_BASE_URL,
-    model: envValue('OLLAMA_MODEL', env) ?? OLLAMA_MODEL,
     temperature: 0
-  } satisfies AiConfig;
+  };
 }
 
 export function isAiConfigured(options?: GetAiConfigOptions) {
@@ -256,19 +132,12 @@ export async function runAiAnalysis(prompt: string, systemPrompt?: string, optio
     ? getExtractionAiConfig(options)
     : getReportAiConfig(options);
 
-  if (workload === 'report' && !config.apiKey) {
+  if (!config.apiKey) {
-    return {
+    throw new Error('ZHIPU_API_KEY is required for AI workloads');
-      provider: 'local-fallback',
-      model: config.model,
-      text: fallbackResponse(prompt)
-    };
   }
 
   const createModel = options?.createModel ?? defaultCreateModel;
   const generate = options?.generate ?? defaultGenerate;
-  const warn = options?.warn ?? console.warn;
-
-  try {
   const model = createModel(config);
 
   const result = await generate({
@@ -281,14 +150,6 @@ export async function runAiAnalysis(prompt: string, systemPrompt?: string, optio
 
   const text = result.text.trim();
   if (!text) {
-      if (workload === 'extraction') {
-        return {
-          provider: 'local-fallback',
-          model: config.model,
-          text: fallbackResponse(prompt)
-        };
-      }
-
     throw new Error('AI SDK returned an empty response');
   }
 
@@ -297,29 +158,6 @@ export async function runAiAnalysis(prompt: string, systemPrompt?: string, optio
     model: config.model,
     text
   };
-  } catch (error) {
-    if (workload === 'report' && shouldFallbackReportError(error)) {
-      warn(`[AI SDK] Report fallback activated: ${asErrorMessage(error)}`);
-
-      return {
-        provider: 'local-fallback',
-        model: config.model,
-        text: fallbackResponse(prompt)
-      };
-    }
-
-    if (workload === 'extraction') {
-      warn(`[AI SDK] Extraction fallback activated: ${asErrorMessage(error)}`);
-
-      return {
-        provider: 'local-fallback',
-        model: config.model,
-        text: fallbackResponse(prompt)
-      };
-    }
-
-    throw error;
-  }
 }
 
 export function __resetAiWarningsForTests() {

lib/server/api/filing-redaction.test.ts

@@ -32,8 +32,8 @@ function filingWithExtraction(): Filing {
         confidence: 0.4
       },
       extractionMeta: {
-        provider: 'ollama',
+        provider: 'zhipu',
-        model: 'qwen3:8b',
+        model: 'glm-4.7-flashx',
         source: 'primary_document',
         generatedAt: '2026-02-01T00:00:00.000Z'
       }

lib/server/task-processors.ts

@@ -689,27 +689,23 @@ async function processAnalyzeFiling(task: Task) {
     source: 'metadata_fallback',
     generatedAt: new Date().toISOString()
   };
+  let filingDocument: Awaited<ReturnType<typeof fetchPrimaryFilingText>> | null = null;
 
   try {
     await setProjectionStage(task, 'analyze.fetch_document', 'Fetching primary filing document');
-    const filingDocument = await fetchPrimaryFilingText({
+    filingDocument = await fetchPrimaryFilingText({
       filingUrl: filing.filing_url,
       cik: filing.cik,
       accessionNumber: filing.accession_number,
       primaryDocument: filing.primary_document ?? null
     });
+  } catch {
+    filingDocument = null;
+  }
 
   if (filingDocument?.text) {
     await setProjectionStage(task, 'analyze.extract', 'Generating extraction context from filing text');
     const ruleBasedExtraction = buildRuleBasedExtraction(filing, filingDocument.text);
-      extraction = ruleBasedExtraction;
-      extractionMeta = {
-        provider: 'deterministic-fallback',
-        model: 'filing-rule-based',
-        source: filingDocument.source,
-        generatedAt: new Date().toISOString()
-      };
-
     const extractionResult = await runAiAnalysis(
       extractionPrompt(filing, filingDocument.text),
       'Return strict JSON only.',
@@ -717,25 +713,18 @@ async function processAnalyzeFiling(task: Task) {
     );
 
     const parsed = parseExtractionPayload(extractionResult.text);
-      if (parsed) {
+    if (!parsed) {
+      throw new Error('Extraction output invalid JSON schema');
+    }
+
     extraction = mergeExtractionWithFallback(parsed, ruleBasedExtraction);
     extractionMeta = {
-          provider: extractionResult.provider === 'local-fallback' ? 'deterministic-fallback' : 'ollama',
+      provider: 'zhipu',
       model: extractionResult.model,
       source: filingDocument.source,
       generatedAt: new Date().toISOString()
     };
   }
-    }
-  } catch {
-    extraction = defaultExtraction;
-    extractionMeta = {
-      provider: 'deterministic-fallback',
-      model: 'metadata-fallback',
-      source: 'metadata_fallback',
-      generatedAt: new Date().toISOString()
-    };
-  }
 
   await setProjectionStage(task, 'analyze.generate_report', 'Generating final filing analysis report');
   const analysis = await runAiAnalysis(

package.json

@@ -16,7 +16,6 @@
     "test:e2e:workflow": "RUN_TASK_WORKFLOW_E2E=1 bun test lib/server/api/task-workflow-hybrid.e2e.test.ts"
   },
   "dependencies": {
-    "@ai-sdk/openai": "^2.0.62",
     "@elysiajs/eden": "^1.4.8",
     "@libsql/client": "^0.17.0",
     "@tailwindcss/postcss": "^4.2.1",