fix(gemini): send the lowest thinking level when thinking is 'none' on Gemini 3 by waleedlatif1 · Pull Request #9278 · simstudioai/sim · GitHub
Skip to content

fix(gemini): send the lowest thinking level when thinking is 'none' on Gemini 3 - #9278

Merged
waleedlatif1 merged 2 commits into
stagingfrom
fix/gemini3-thinking-none
Oct 21, 2026
Merged

fix(gemini): send the lowest thinking level when thinking is 'none' on Gemini 3#9278
waleedlatif1 merged 2 commits into
stagingfrom
fix/gemini3-thinking-none

Conversation

@waleedlatif1

@waleedlatif1 waleedlatif1 commented Oct 21, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • Thinking Level none on Gemini 3 models sent no thinkingConfig, so the API fell back to the model's default level (medium/high) and kept thinking
  • Gemini 3 has no off switch, so none now sends the lowest level the model declares: minimal where offered (closest to off per Google's docs), low on models that reject minimal (3.7/3.8-flash, 3.1/3-pro)
  • Uncataloged Gemini 3 models and Gemini 2.5 behavior are unchanged
  • Updated the noneMode TSDoc to reflect the Gemini 3 rule

Type of Change

  • Bug fix

Testing

  • New regression test in providers/gemini/core.test.ts (3.6-flash → MINIMAL, 3.8-flash → LOW); failed before the fix, passes after
  • vitest run providers/ — 177 files, 2103 tests passing; biome clean
  • Live API check through executeGeminiRequest with none (output tokens incl. thoughts, before → after): 3.6-flash 418 → 20, 3.8-flash 277 → 21, 3.7-flash 325 → 20, 3.5-flash 518 → 89; 3.1-pro-preview still thinks (it only runs in thinking mode)
  • Probed the raw API: MINIMAL is rejected on 3.7/3.8-flash and 3.1-pro; thinkingBudget: 0 is rejected on 3.5-flash-lite and 3.1-pro, so it isn't a safe general off switch

Checklist

  • Code follows project style guidelines
  • Self-reviewed my changes
  • Tests added/updated and passing (new tests pass the test-audit authoring gate)
  • No new warnings introduced
  • I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

@vercel

vercel Bot commented Oct 21, 2026 •
edited
Loading

Copy link
Copy Markdown

cubic-dev-ai[bot]
cubic-dev-ai Bot reviewed Oct 21, 2026
View reviewed changes

@cubic-dev-ai cubic-dev-ai Bot left a comment •
edited
Loading

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

All reported issues were addressed across 3 files

Heads up: you’re close to your flex budget. Increase your flex budget so reviews don’t pause.

Turn on auto-fix | Re-trigger cubic

Comment thread apps/sim/providers/gemini/core.ts Outdated
@greptile-apps

greptile-apps Bot commented Oct 21, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

RetriggerView in GreptileConfidence Score: 5/5

[Medium impact] The PR appears safe to merge; no actionable issues were found.

Summary

Gemini 3 requests with thinkingLevel: 'none' now send the lowest thinking level declared in the model catalog.

  • Gemini 3 requests use the lowest listed level when thinking is set to none.

Reviews (3) · Last reviewed commit: "fix(gemini): resolve the lowest thinking..." · Reviewed by Greptile

@waleedlatif1

waleedlatif1 commented Oct 21, 2026

Copy link
Copy Markdown
Collaborator Author

@greptile

@waleedlatif1

waleedlatif1 commented Oct 21, 2026

Copy link
Copy Markdown
Collaborator Author

@cubic-dev-ai review this PR

@cubic-dev-ai

cubic-dev-ai Bot commented Oct 21, 2026

Copy link
Copy Markdown
Contributor

@cubic-dev-ai review this PR

@waleedlatif1 I have started the AI code review. It will take a few minutes to complete.

cubic-dev-ai[bot]
cubic-dev-ai Bot reviewed Oct 21, 2026
View reviewed changes

@cubic-dev-ai cubic-dev-ai Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No issues found across 4 files

Confidence score: 5/5

  • Automated review surfaced no issues in the provided summaries.
  • No files require special attention.

Heads up: you’re close to your flex budget. Increase your flex budget so reviews don’t pause.

Turn on auto-fix | Re-trigger cubic

@waleedlatif1

waleedlatif1 commented Oct 21, 2026

Copy link
Copy Markdown
Collaborator Author

@waleedlatif1
waleedlatif1 merged commit daf2379 into staging Oct 21, 2026
55 checks passed
@waleedlatif1
waleedlatif1 deleted the fix/gemini3-thinking-none branch October 21, 2026 16:37
@waleedlatif1 waleedlatif1 mentioned this pull request Oct 21, 2026
Merged

This branch was previously deployed

1 inactive deployment
Preview — ab33c5da Deployed Oct 21, 2026 by vercel[bot]
View deployment
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@cubic-dev-ai cubic-dev-ai[bot] cubic-dev-ai[bot] left review comments

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@waleedlatif1