feat(openai): add GPT-5.6 family (sol, terra, luna) support#876
Conversation
Add the GPT-5.6 model family to both the OpenAI Codex (ChatGPT subscription) and OpenAI Native (API) providers: - gpt-5.6-sol: flagship frontier reasoning/coding model - gpt-5.6-terra: balanced everyday model - gpt-5.6-luna: fastest, most affordable family member Update the default model id for both providers to gpt-5.6-sol and update the corresponding unit test assertions. Closes #871
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
📝 WalkthroughWalkthroughOpenAI native and Codex providers now support three GPT-5.6 models and select ChangesOpenAI GPT-5.6 support
Estimated code review effort: 2 (Simple) | ~10 minutes Possibly related PRs
Suggested labels: Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
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 |
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
taltas
left a comment
There was a problem hiding this comment.
One possible tweak to context window
| export const openAiCodexModels = { | ||
| "gpt-5.6-sol": { | ||
| maxTokens: 128000, | ||
| contextWindow: 400000, |
There was a problem hiding this comment.
These context windows are not accurate, the rest of the numbers right, but I would double check. Maybe I have interpreted things wrong.
Address review feedback on PR #876: the GPT-5.6 model context windows in the Codex provider were set to 400000, but the OpenAI codex models.json shows context_window and max_context_window of 372000 for all three GPT-5.6 variants (sol, terra, luna). Updated accordingly.
|
Luna is currently broken in from codex... not much we can do, its on openai's end |
taltas
left a comment
There was a problem hiding this comment.
Tested, works except Luna on codex, which is an OpenAi problem, not ours.
Related GitHub Issue
Closes: #871
Description
This PR adds support for the GPT-5.6 model family from OpenAI, as requested in the linked issue. The three new models —
gpt-5.6-sol,gpt-5.6-terra, andgpt-5.6-luna— are registered in both OpenAI provider paths:packages/types/src/providers/openai-codex.ts): The ChatGPT Plus/Pro subscription path. The three models are added with subscription-based pricing (input/output cost0), a 400k context window, image support, prompt caching, and the full reasoning-effort ladder (none→max). The default model id is updated fromgpt-5.3-codextogpt-5.6-sol.packages/types/src/providers/openai.ts): The direct API path. The three models are added with per-token pricing, cache write/read pricing, long-context pricing tiers, and service-tier definitions (flex/priority).gpt-5.6-solandgpt-5.6-terrause a 1.05M context window whilegpt-5.6-lunauses 400k. The default model id is updated fromgpt-5.1-codex-maxtogpt-5.6-sol.The default-model assertions in the existing unit tests (
openai-codex.spec.tsandopenai-native.spec.ts) were updated to match the new defaults. No other references to the old default ids required changes — remaining mentions ofgpt-5.3-codex/gpt-5.1-codex-maxare legitimate model definitions and dedicated test cases for those still-supported models.Test Procedure
srcworkspace, ran the affected provider test suites:turbo runbuild) passed on push.feat/gpt-5.6-family-supportand re-running the command above.Pre-Submission Checklist
Screenshots / Videos
Not applicable — this change adds provider model definitions and updates test assertions; no UI changes are involved.
Documentation Updates
Additional Notes
Model capability metadata (context windows, pricing, reasoning-effort ladders, service tiers) was modeled on the existing GPT-5.x entries and the OpenAI model documentation linked in the issue. The Codex (subscription) variants keep
inputPrice/outputPriceat0to reflect subscription-based access, while the Native (API) variants carry per-token pricing consistent with OpenAI's published rates.Get in Touch
Summary by CodeRabbit