feat(plugin): surface reasoning-effort variants from /v1/model/info reported by LiteLLM

[JCHacking]

Summary

Read per-model reasoning effort support from LiteLLM's /v1/model/info and expose each supported level as an OpenCode variant with reasoningEffort set.

Type of change

• 🐛 Bug fix (non-breaking)
• ✨ New feature (non-breaking)
• 💥 Breaking change
• 📝 Documentation only
• 🔧 Internal / refactor

Checklist

• npm run typecheck passes
• No new runtime dependencies (or justified in this PR description)
• README updated if public API or behavior changed
• CHANGELOG.md updated under ## [Unreleased]
• Commit messages follow Conventional Commits

How was this tested?

I tested a model that had other variants that didn't include OpenCode by default, and with this change, it now reads them correctly.

Screenshots / logs (optional)

Summary by CodeRabbit

• New Features

  • Added support for exposing LiteLLM-reported reasoning-effort model variants.
  • Models supporting low, medium, or high reasoning effort can now be selected with the corresponding setting.
  • Reasoning-effort capabilities are automatically detected from LiteLLM model metadata.

• Documentation

  • Updated the changelog and model documentation with details about reasoning-effort variants.

[coderabbitai]

📝 Walkthrough

Walkthrough

The plugin now fetches LiteLLM model metadata, detects supported reasoning-effort levels, and exposes corresponding OpenCode model variants with reasoningEffort settings. Documentation and the changelog describe the new behavior.

Changes

Reasoning-effort model variants

Layer / File(s)	Summary
Capability metadata detection src/types/index.ts, src/utils/litellm-api.ts	LiteLLMModelInfo stores supported reasoning efforts. Discovery detects supports_<name>_reasoning_effort flags from model metadata and LiteLLM parameters.
Model metadata discovery src/plugin/discover.ts	Model listing and /v1/model/info retrieval run concurrently. Model-list failures stop discovery. Metadata failures are logged, and models continue without metadata.
Variant generation and documentation src/plugin/build-model.ts, src/plugin/index.ts, README.md, CHANGELOG.md	Model conversion creates variants for each supported reasoning effort and assigns the matching reasoningEffort value. Documentation describes the generated variants.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Sequence Diagram(s)

sequenceDiagram
  participant ModelDiscovery
  participant LiteLLM
  participant ModelConverter
  participant OpenCodeConfig
  ModelDiscovery->>LiteLLM: Fetch model list and /v1/model/info
  LiteLLM-->>ModelDiscovery: Return models and optional metadata
  ModelDiscovery->>ModelConverter: Pass matched model metadata
  ModelConverter->>OpenCodeConfig: Create reasoning-effort variants

Loading

Possibly related issues

• Read variants per model #13 — Requests per-model reasoning variants from supports_*_reasoning_effort flags.

Possibly related PRs

• yuseferi/opencode-litellm#10 — Both changes update LiteLLM model-info discovery and model construction.
• yuseferi/opencode-litellm#11 — Both changes pass LiteLLM metadata through discovery and model conversion.

Suggested reviewers: yuseferi, overbit

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 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 exposing LiteLLM-reported reasoning-effort variants from the /v1/model/info endpoint.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
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

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

Warning

⚠️ This pull request shows signs of AI-generated slop (redundant_comments). It has been flagged by CodeRabbit slop detection and should be reviewed carefully.

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@src/plugin/discover.ts`:
- Around line 193-194: Update the model construction flow around buildModelV2 to
merge the /v1/model/info metadata from infoByName into each /v1/models record
before building it, using the same field-precedence merge established in
src/plugin/index.ts. Pass the merged model so capabilities, limits, mode, and
reasoning-effort flags are reflected correctly while preserving the existing
resolvedApi, perModelApi, and model identity inputs.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 4b9ed210-d541-48f2-b86c-23a5932ef8e0

📥 Commits

Reviewing files that changed from the base of the PR and between baa8874 and 938f239.

📒 Files selected for processing (7)

• CHANGELOG.md
• README.md
• src/plugin/build-model.ts
• src/plugin/discover.ts
• src/plugin/index.ts
• src/types/index.ts
• src/utils/litellm-api.ts