fix(plugin): map real pricing from /v1/model/info into cost field

[zyc]

Summary

Discovered models never got a cost — toConfigModel() reads /v1/model/info for mode, token limits, and capability flags, but drops input_cost_per_token/output_cost_per_token on the floor. Every model lands in opencode.json without a cost field, so OpenCode's own default (0) makes every call look free in the picker and in /cost, regardless of what the proxy actually bills.

Type of change

• 🐛 Bug fix (non-breaking)

Checklist

• npm run typecheck passes
• No new runtime dependencies
• README updated (new "Real pricing" row in the features table, "How it works" step 4)
• CHANGELOG.md updated under ## [Unreleased]
• Commit messages follow Conventional Commits

How was this tested?

Live, against a self-hosted LiteLLM 1.92.0 proxy (21 models, 15 chat + 6 non-chat) fronting real vLLM/NIM backends, with OpenCode 1.18.13.

• opencode models <provider> --verbose: all models that have a price in the proxy's model_list now carry the matching cost in the discovered entry (USD/token from /v1/model/info converted to USD/1M-token, OpenCode's convention). Models the proxy has no price for are left without a cost field — not defaulted to 0 — same as before this fix, so nothing is falsely marked free.
• A real chat completion's exported session cost matched prompt_tokens * input_cost_per_token + completion_tokens * output_cost_per_token computed from the proxy's own price table, to the fraction of a cent, across two independent runs.

Screenshots / logs (optional)

Before (unpatched):

"cost": { "input": 0, "output": 0, "cache": { "read": 0, "write": 0 } }

After (this PR), same model:

"cost": { "input": 0.07, "output": 0.07, "cache": { "read": 0, "write": 0 } }

matching the proxy's configured input_cost_per_token: 0.00000007 / output_cost_per_token: 0.00000007 exactly (×1e6).

Summary by CodeRabbit

• New Features

  • Added real input and output token pricing for discovered models.
  • Pricing is converted to per-million-token costs for display.
  • Models without available pricing remain unpriced.

• Documentation

  • Updated the README and changelog to describe pricing enrichment and handling of unavailable pricing.

[coderabbitai]

📝 Walkthrough

Walkthrough

The change adds optional LiteLLM input and output pricing fields, converts available per-token prices to OpenCode’s per-million-token cost fields, omits costs when pricing is unavailable, and documents the behavior.

Changes

Pricing propagation

Layer / File(s)	Summary
Pricing contract and enrichment src/types/index.ts, src/plugin/index.ts	LiteLLM model types now include optional input and output token prices. Model enrichment copies these values from LiteLLM metadata.
Cost mapping and documentation src/plugin/index.ts, README.md, CHANGELOG.md	Generated OpenCode models conditionally include converted per-million-token costs. The README and changelog document the pricing behavior.

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

Sequence Diagram(s)

sequenceDiagram
  participant LiteLLM as LiteLLM /v1/model/info
  participant Enrichment as Model enrichment
  participant OpenCode as OpenCode model configuration
  LiteLLM->>Enrichment: input and output cost per token
  Enrichment->>OpenCode: converted per-million-token cost

Loading

Possibly related PRs

• yuseferi/opencode-litellm#10: Extends the model-discovery enrichment used by this pricing change.
• yuseferi/opencode-litellm#11: Also modifies LiteLLM pricing types and OpenCode cost conversion.
• yuseferi/opencode-litellm#14: Extends LiteLLM /v1/model/info enrichment in the same implementation areas.

Suggested reviewers: yuseferi

🚥 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 mapping LiteLLM pricing into the OpenCode cost field.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
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.}

[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/index.ts`:
- Around line 123-132: The cost mapping around LiteLLMModel pricing must not
represent a missing side as zero. Update the condition and cost construction so
cost is emitted only when both input_cost_per_token and output_cost_per_token
are present, unless the target schema explicitly supports partial cost objects;
preserve reported values and add fixtures covering input-only and output-only
pricing.

🪄 Autofix

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: 2b5d7541-45c2-4e94-9309-de51d42a4408

📥 Commits

Reviewing files that changed from the base of the PR and between baa8874 and 218082c.

📒 Files selected for processing (4)

• CHANGELOG.md
• README.md
• src/plugin/index.ts
• src/types/index.ts