fix(config): support QWEN_CODE_API_TIMEOUT_MS across OAuth and non-OAuth paths by B-A-M-N · Pull Request #3629 · QwenLM/qwen-code

B-A-M-N · 2026-04-26T00:57:17Z

Summary

Adds QWEN_CODE_API_TIMEOUT_MS as an environment override for model generation timeout.

This is intended for slow local or OpenAI-compatible backends where editing settings.json is inconvenient.

What changed

Supports QWEN_CODE_API_TIMEOUT_MS timeout override
Preserves precedence:

modelProvider > env var > settings > default
Applies the override to both:
- standard/non-OAuth config resolution
- AuthType.QWEN_OAUTH config resolution
Floors float values to integer milliseconds
Ignores invalid, zero, negative, or non-numeric values
Keeps default behavior unchanged when the env var is not set

Review feedback addressed

This updates the original implementation based on review feedback:

Fixed the OAuth-path bug where resolveModelConfig() returned early before the env override was applied
Added Qwen OAuth coverage so QWEN_CODE_API_TIMEOUT_MS is tested against the default auth path
Normalized timeout parsing with integer millisecond handling
Added edge-case coverage for invalid values, negative values, whitespace-padded values, large values, and precedence behavior

Validation

npx vitest run packages/core/src/models/modelConfigResolver.test.ts
npm run lint
full relevant test coverage passes locally

Scope / risk

Low risk. This only changes timeout resolution when QWEN_CODE_API_TIMEOUT_MS is explicitly provided.

No breaking changes.

Adds support for QWEN_CODE_API_TIMEOUT_MS as an environment override for model generation timeout. Qwen Code already supports timeout configuration via: settings.model.generationConfig.timeout This change introduces an env-based override for users running slow local/OpenAI-compatible backends where editing config is less convenient. Precedence: modelProvider > env var > settings > default (120000ms) Behavior: - Valid positive env values override configured timeout - Invalid values are ignored - Default behavior remains unchanged (applied in buildClient()) Note: The 5-minute timeout reported in QwenLM#1045 originally came from undici's default bodyTimeout, which is now disabled (bodyTimeout:0). The modelConfigResolver default is 120000ms (2 minutes). Includes unit tests covering precedence and validation. Closes QwenLM#1045 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

wenshao

The PR successfully implements the QWEN_CODE_API_TIMEOUT_MS environment variable override for most authentication types. However, there are a few minor items to address for consistency and robustness, particularly regarding the QWEN_OAUTH code path.\n\n_— Gemini CLI /review_

wenshao · 2026-04-26T01:30:58Z

    sources,
  );

+  // ---- Env override: QWEN_CODE_API_TIMEOUT_MS ----


[Suggestion] The QWEN_CODE_API_TIMEOUT_MS override logic here is skipped for QWEN_OAUTH due to the early return at the start of resolveModelConfig. Consider moving this logic or applying it within resolveQwenOAuthConfig for consistency across all auth types.\n\n_— Gemini CLI /review_

wenshao · 2026-04-26T01:30:58Z

+    const envTimeout = env['QWEN_CODE_API_TIMEOUT_MS'];
+    if (envTimeout !== undefined) {
+      const parsed = Number(envTimeout);
+      if (Number.isFinite(parsed) && parsed > 0) {


[Nice to have] Use Math.floor(Number(envTimeout)) or parseInt(envTimeout, 10) to ensure the timeout value is an integer, which is the standard format for millisecond durations.\n\n_— Gemini CLI /review_

wenshao · 2026-04-26T01:30:58Z

        expect(result.config.timeout).toBe(60000);
        expect(result.sources['timeout'].kind).toBe('modelProviders');
      });
+


[Suggestion] Add a test case to verify that QWEN_CODE_API_TIMEOUT_MS correctly overrides settings when authType is AuthType.QWEN_OAUTH. This would have helped catch the missing support in the implementation.\n\n_— Gemini CLI /review_

wenshao · 2026-04-26T01:51:35Z

@@ -247,6 +247,24 @@ export function resolveModelConfig(
    sources,


[Critical] The timeout env override is only applied in the non-OAuth path because resolveModelConfig() returns early for AuthType.QWEN_OAUTH before reaching this new block. resolveQwenOAuthConfig() never reads input.env, so Qwen OAuth users still cannot override API timeouts with QWEN_CODE_API_TIMEOUT_MS.

Please move this override into shared generation-config resolution used by both paths, preserving modelProvider > env > settings, and add a Qwen OAuth test for the env timeout override.

— gpt-5.5 via Qwen Code /review

wenshao · 2026-04-26T05:06:58Z

    sources,
  );

+  // ---- Env override: QWEN_CODE_API_TIMEOUT_MS ----


[Critical] This override is applied only after the AuthType.QWEN_OAUTH early return above, so QWEN_CODE_API_TIMEOUT_MS is ignored for Qwen OAuth sessions. That means the new env override does not work on the default Qwen auth path, even though the downstream provider can consume config.timeout.

Please apply the same env-timeout merge in resolveQwenOAuthConfig() as well, or factor this logic into a shared helper used by both resolver paths after resolveGenerationConfig(), while preserving modelProvider > env > settings precedence.

— gpt-5.5 via Qwen Code /review

wenshao

No issues found. LGTM! ✅ — gpt-5.5 via Qwen Code /review

Covers: large timeout values, whitespace-padded env values, negative env values, and reinforces provider > env > settings precedence. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Adds support for QWEN_CODE_API_TIMEOUT_MS as an environment override for model generation timeout. Closes QwenLM#13 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

wenshao · 2026-04-26T16:11:17Z

+  // Regression: when settings.model.name is set but doesn't match any
+  // provider, the code should NOT fall through to OPENAI_MODEL/QWEN_MODEL.
+  // The env-matched provider should only supply metadata when no model is requested.
+  const mockSettings = makeMockSettings({


[Critical] The new regression test is outside the describe('modelConfigUtils', ...) block where makeMockSettings is scoped. This makes npm run typecheck fail with TS2552: Cannot find name 'makeMockSettings', and the focused vitest run fails with ReferenceError: makeMockSettings is not defined. Move this test inside the existing describe('resolveCliGenerationConfig', ...) block, or hoist makeMockSettings to module scope.

— gpt-5.5 via Qwen Code /review

B-A-M-N · 2026-04-26T21:40:57Z

Follow-up note: I've created a refactor PR (#3651, now closed) to dedupe the duplicated QWEN_CODE_API_TIMEOUT_MS logic into a shared applyTimeoutEnvOverride() helper. Realized this should be recreated after this PR merges to main rather than being based on the feature branch. Will open a clean refactor PR once #3629 lands.

wenshao

No issues found. LGTM! ✅ — gpt-5.5 via Qwen Code /review

…Override helper Deduplicates the timeout env override logic that was duplicated in resolveModelConfig() and resolveQwenOAuthConfig() after PR QwenLM#3629 merged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…uth paths (QwenLM#3629) * feat(config): support API timeout env override Adds support for QWEN_CODE_API_TIMEOUT_MS as an environment override for model generation timeout. Qwen Code already supports timeout configuration via: settings.model.generationConfig.timeout This change introduces an env-based override for users running slow local/OpenAI-compatible backends where editing config is less convenient. Precedence: modelProvider > env var > settings > default (120000ms) Behavior: - Valid positive env values override configured timeout - Invalid values are ignored - Default behavior remains unchanged (applied in buildClient()) Note: The 5-minute timeout reported in QwenLM#1045 originally came from undici's default bodyTimeout, which is now disabled (bodyTimeout:0). The modelConfigResolver default is 120000ms (2 minutes). Includes unit tests covering precedence and validation. Closes QwenLM#1045 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test(core): add edge-case tests for QWEN_CODE_API_TIMEOUT_MS Covers: large timeout values, whitespace-padded env values, negative env values, and reinforces provider > env > settings precedence. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(config): support QWEN_CODE_API_TIMEOUT_MS override Adds support for QWEN_CODE_API_TIMEOUT_MS as an environment override for model generation timeout. Closes QwenLM#13 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>

github-actions Bot mentioned this pull request Apr 26, 2026

📊 AI CLI 工具社区动态日报 2026-04-26 gsscsd/big_model_radar#245

Open

wenshao reviewed Apr 26, 2026

View reviewed changes

wenshao requested changes Apr 26, 2026

View reviewed changes

github-actions Bot mentioned this pull request Apr 26, 2026

📊 AI CLI 工具社区动态日报 2026-04-26 borq168/radar-forge#27

Open

Huynhthuongg mentioned this pull request Apr 26, 2026

# AI CLI 工具社区动态日报 2026-04-26 Huynhthuongg/Claude-OpenAI#7

Open

wenshao requested changes Apr 26, 2026

View reviewed changes

wenshao mentioned this pull request Apr 26, 2026

feat(stats): add model cost estimation + fix model priority #3631

Closed

wenshao previously approved these changes Apr 26, 2026

View reviewed changes

B-A-M-N dismissed wenshao’s stale review via 79eae30 April 26, 2026 15:14

B-A-M-N force-pushed the fix/configurable-api-timeout branch from 79eae30 to 005d822 Compare April 26, 2026 15:37

test(core): add edge-case tests for QWEN_CODE_API_TIMEOUT_MS

cc47043

Covers: large timeout values, whitespace-padded env values, negative env values, and reinforces provider > env > settings precedence. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

B-A-M-N force-pushed the fix/configurable-api-timeout branch from 005d822 to 0a5b96d Compare April 26, 2026 15:46

feat(config): support QWEN_CODE_API_TIMEOUT_MS override

5eb2406

Adds support for QWEN_CODE_API_TIMEOUT_MS as an environment override for model generation timeout. Closes QwenLM#13 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

B-A-M-N force-pushed the fix/configurable-api-timeout branch from 0a5b96d to 5eb2406 Compare April 26, 2026 15:58

wenshao requested changes Apr 26, 2026

View reviewed changes

B-A-M-N changed the title ~~feat(config): support API timeout env override~~ fix(config): support QWEN_CODE_API_TIMEOUT_MS across OAuth and non-OAuth paths Apr 26, 2026

B-A-M-N mentioned this pull request Apr 26, 2026

refactor(config): dedupe QWEN_CODE_API_TIMEOUT_MS env override logic #3651

Closed

B-A-M-N mentioned this pull request Apr 26, 2026

feat(stats): add optional cost estimation to /stats model B-A-M-N/qwen-code#2

Closed

3 tasks

wenshao approved these changes Apr 26, 2026

View reviewed changes

wenshao merged commit 70127b5 into QwenLM:main Apr 26, 2026
24 of 25 checks passed

B-A-M-N mentioned this pull request Apr 26, 2026

refactor(config): dedupe QWEN_CODE_API_TIMEOUT_MS env override logic #3653

Merged

B-A-M-N mentioned this pull request Apr 27, 2026

refactor(core): dedupe QWEN_CODE_API_TIMEOUT_MS env override logic B-A-M-N/qwen-code#3

Closed

1 task

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix(config): support QWEN_CODE_API_TIMEOUT_MS across OAuth and non-OAuth paths#3629

fix(config): support QWEN_CODE_API_TIMEOUT_MS across OAuth and non-OAuth paths#3629
wenshao merged 3 commits into
QwenLM:mainfrom
B-A-M-N:fix/configurable-api-timeout

B-A-M-N commented Apr 26, 2026 •

edited

Loading

Uh oh!

wenshao left a comment •

edited

Loading

Uh oh!

wenshao Apr 26, 2026 •

edited

Loading

Uh oh!

wenshao Apr 26, 2026 •

edited

Loading

Uh oh!

wenshao Apr 26, 2026 •

edited

Loading

Uh oh!

wenshao Apr 26, 2026

Uh oh!

wenshao Apr 26, 2026

Uh oh!

wenshao left a comment

Uh oh!

wenshao Apr 26, 2026

Uh oh!

B-A-M-N commented Apr 26, 2026

Uh oh!

wenshao left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		@@ -247,6 +247,24 @@ export function resolveModelConfig(
		sources,

Conversation

B-A-M-N commented Apr 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

What changed

Review feedback addressed

Validation

Scope / risk

Uh oh!

wenshao left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wenshao Apr 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wenshao Apr 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wenshao Apr 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wenshao Apr 26, 2026

Choose a reason for hiding this comment

Uh oh!

wenshao Apr 26, 2026

Choose a reason for hiding this comment

Uh oh!

wenshao left a comment

Choose a reason for hiding this comment

Uh oh!

wenshao Apr 26, 2026

Choose a reason for hiding this comment

Uh oh!

B-A-M-N commented Apr 26, 2026

Uh oh!

wenshao left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

B-A-M-N commented Apr 26, 2026 •

edited

Loading

wenshao left a comment •

edited

Loading

wenshao Apr 26, 2026 •

edited

Loading

wenshao Apr 26, 2026 •

edited

Loading

wenshao Apr 26, 2026 •

edited

Loading