fix(google): send thinkingLevel for any model with an effort ladder

[DevMello]

Summary

• Selecting a reasoning effort for gemini-3.1-pro-preview now actually sends generationConfig.thinkingConfig.thinkingLevel on the wire. The catalog has advertised low/medium/high for this model since the 3.6 rollout, but the adapter hardcoded thinking to gemini-3.5-flash and gemini-3.6-flash, so the picker choice was silently ignored.
• The gate is now ladder driven: any model whose provider config advertises reasoning efforts (registry preset or user config) gets the mapped level. The old flash slice stays hardcoded so unladdered configs keep their current behavior.
• Vertex participates only when a user explicitly configures a ladder. The seed google-vertex entry ships none, so the Vertex freeze documented in the 260722 rollout plan is preserved. Image models are excluded because thinkingConfig would suppress responseModalities.

Verification

• bun test tests/google-hardening.test.ts tests/google-adapter.test.ts tests/google-antigravity-wire.test.ts tests/google-vertex-stream.test.ts (77 pass)
• bun run typecheck, bun run lint:gui, bun run privacy:scan
• bun test --isolate ./tests/ full suite green (7999 pass, 0 fail)
• New tests: pro-preview ladder emits the level and clamps minimal to low, Vertex emits only with an explicit ladder, unladdered direct flash keeps its slice, image models keep responseModalities under a provider-wide ladder.

Checklist

• Scope stays focused and avoids unrelated cleanup.
• Docs or release notes were updated when needed.
• Security-sensitive changes were reviewed for secrets, auth, and unsafe defaults.

Summary by CodeRabbit

• New Features
  • Google AI models now support configured reasoning levels where available.
  • Reasoning settings are automatically adjusted to each model’s supported capabilities.
  • Existing image-generation and fallback behavior remains supported.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: a0e68511-b96c-48a2-bb53-6f0a22d2983a

📥 Commits

Reviewing files that changed from the base of the PR and between e3ba14f and 753ffdc.

📒 Files selected for processing (2)

• src/adapters/google.ts
• tests/google-hardening.test.ts

---

📝 Walkthrough

Walkthrough

The Google adapter now uses configured reasoning-effort ladders for eligible Gemini and Vertex models. Tests cover effort mapping, clamping, Vertex configuration, Flash fallback, and image-model exclusions.

Changes

Google reasoning request handling

Layer / File(s)	Summary
Reasoning-aware request construction src/adapters/google.ts	The adapter applies configured reasoning efforts to eligible Gemini and Vertex requests. It preserves direct Flash fallback behavior and excludes Cloud Code Assist and image-capable models.
Model-specific behavior validation tests/google-hardening.test.ts	Tests cover Pro preview models, Vertex ladders, unsupported-effort clamping, direct Flash fallback, omitted thinking configuration, and image response modalities.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• lidge-jun/opencodex#309: Both modify Google thinkingConfig.thinkingLevel request handling.
• lidge-jun/opencodex#355: Both test Google reasoning and image-modality request behavior.
• lidge-jun/opencodex#494: Both modify adapter reasoning-effort handling and related tests.

Suggested reviewers: ingwannu, wibias, lidge-jun

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: sending thinkingLevel for Google models with configured reasoning ladders.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}