Repository navigation
Conversation
OpenCode Go serves grok-4.7 only on /v1/responses, but the provider sent it to Chat Completions with generic limits. Add it to the Responses format set with a native configuration from xAI's model page (500K context, text and image input, reasoning effort low/medium/high/xhigh with high as the default) and the Go pricing with its 200K tier and no cache-write rate. Closes Zoo-Code-Org#1996
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info
📜 Recent review details
📝 Summary
Merge Risk: ⚪ Minimal · up to This adds Grok 4.7 support for OpenCode Go with the specified routing, limits, and pricing. No merge-blocking risk was identified; the only caveat is that no authenticated gateway request was run. Pre-merge checks |
 |
| Check name | Status | Explanation |
|---|---|---|
| Linked Issues check |  |
Issue #1996 coding requirements are met. The PR adds grok-4.7 to OPENCODE_GO_RESPONSES_FORMAT_MODELS and adds native metadata with a 500,000-token context, image support, prompt-cache support, rea… |
| Out of Scope Changes check | |
The changed files stay within issue #1996. Registry changes implement routing and native model configuration. Provider tests verify the requested request paths, capabilities, pricing, and usage behavi… |
| Regression Evidence | |
Focused coverage exists for each changed behavior. The registry tests verify Grok 4.7 metadata, Responses classification, and the intentionally unset cache-write price (packages/types/src/tests/op… |
| Security Boundaries | |
No changed path meets the security failure conditions. The production diff only adds static grok-4.7 metadata and its exact ID to the Responses allowlist in `packages/types/src/providers/opencode-go… |
| Persistence Integrity | |
PASS — The pull request changes only the OpenCode Go model registry, format classification, and tests. The changed production code adds static grok-4.7 metadata and routing; it does not add or modif… |
| Lifecycle Resource Cleanup | |
No changed lifecycle path can be shown to leak a resource or duplicate work. The PR changes only the Grok 4.7 registry entry and the Responses-format model set in `packages/types/src/providers/opencod… |
| Title check | |
The title clearly and concisely describes the primary change: routing OpenCode Go's Grok 4.7 integration through the Responses API. |
| Description check | |
The description follows the repository template. It links issue #1996, explains the routing and native configuration changes, documents focused and broad test procedures with results, completes the ch… |
✨ Finishing Touches
-
-
-
🧪 Generate unit tests (beta)
-
-
-
- Create a new PR
-
-
-
-
-
- Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts
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.
Comment @coderabbitai help to get the list of available commands.
Review statusThanks for contributing. This comment tracks the review sequence and the next action. Current step: Awaiting fresh human maintainer or CODEOWNER approval. Automated review is complete for the latest commit but does not replace human approval. Review-state labels are managed by this workflow; do not edit them manually. |
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
github-actions
Bot
added
coderabbit-review-active
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
github-actions
Bot
added
awaiting-maintainer
This branch has not been deployed
1 participant
Related GitHub Issue
Closes: #1996
Description
OpenCode Go serves
grok-4.7only on the Responses API, but neither format set listed it, so requests went to Chat Completions with the fetcher's generic limits.grok-4.7joinsOPENCODE_GO_RESPONSES_FORMAT_MODELSexplicitly (no Grok-wide rule, as the issue asks), so streaming requests andcompletePromptboth use/v1/responses.low/medium/high/xhigh, defaulthigh, which the Responses path sends asreasoning.effort.grok-4.6,maxTokensequals the context window; requests still go through the shared max-output rule (100K here).longContextPricingwith a 200K threshold). No cache-write rate is set, since Go doesn't publish one. The Responses path already prices throughcalculateApiCostOpenAI, which applies the tier.Test Procedure
Run from
src/unless noted:pnpm exec vitest run api/providers/__tests__/opencode-go.spec.ts api/providers/fetchers/__tests__/opencode-go.spec.ts: 100 pass. New cases: streaming and non-streaming requests go to Responses withreasoning.effort: "high"and no temperature (never Chat Completions or Messages); tool-call partials, completed tool calls and cached usage are priced at the Go rates; cost at 200,000 vs 200,001 input tokens plus an above-threshold case with cached input and output; and the fetcher resolves an ID-only catalog entry to the native configuration. Before the fix these fail: requests go to Chat Completions and the fetcher returns the generic 32,768-token fallback.packages/types:pnpm exec vitest run src/__tests__/opencode-go.test.ts: 42 pass (registry entry, Responses set, no cache-write rate). Existing Grok 4.5 (Chat Completions), Grok 4.6, GPT and other protocol cases are unchanged and pass.pnpm exec vitest run api: 123 files, 3,061 passed. All ofpackages/types: 31 files, 453 passed.pnpm check-types(insrcandpackages/types) ✅,pnpm lint✅ (pre-commit hook), Prettier ✅.Not run: an authenticated request to the Go gateway. As in the issue, everything is verified with mocked SDK calls.
Pre-Submission Checklist
Visual Snapshots
N/A, no UI change.
Videos (interaction / animation only)
N/A.
Documentation Updates
Additional Notes
#2000 (Claude Haiku 5.5, #1995) touches the same files in different places; the two branches merge cleanly with each other in either order.
Get in Touch
Discord: tosin.py