chore: bump coder/fantasy for gpt-5.6 Responses routing - #27132
Conversation
Pulls in coder/fantasy#46, which routes the gpt-5.6 family (sol, terra, luna) through the OpenAI Responses API instead of Chat Completions, and coder/fantasy#41, which surfaces the Anthropic refusal stop_reason as content-filter.
ibetitsmike
changed the title
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: a68b17a1b9
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| // 16) coder/fantasy#46, route the gpt-5.6 family (sol, terra, luna) | ||
| // through the OpenAI Responses API. |
There was a problem hiding this comment.
This bump explicitly makes gpt-5.6-sol, gpt-5.6-terra, and gpt-5.6-luna usable, but the seeded OpenAI gpt-5 price rows still stop at gpt-5.5-pro (coderd/aibridge/prices/data/prices.json:450-465). In budgeted deployments, any request routed to one of these new model IDs will miss GetAIModelPriceByProviderModel, which resolveTokenUsageCost records as NULL cost, and recordTokenUsageAndSpend then skips the spend update, so usage of the newly enabled models is not counted toward AI budgets or cost reports (coderd/aibridgedserver/cost.go:52-62, coderd/aibridgedserver/aibridgedserver.go:347-349). Please add seed rows for the new model IDs alongside the routing bump.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
ai gateway is not a part of this PR
2 participants
gpt-5.6 models were unusable with agents: fantasy's Responses allowlist did not include the new family, so
IsResponsesModelreturned false and chatd fell back to the Chat Completions path (no reasoning params, no encrypted reasoning continuity).Changes
charm.land/fantasyreplace pin to coder/fantasy6da0c3b10237(coder_2_33), pulling in:gpt-5.6-sol,gpt-5.6-terra,gpt-5.6-lunathrough the OpenAI Responses API as reasoning models.Verification
/v1/responseswithreasoning.effort,include: ["reasoning.encrypted_content"], andstore: falsecompletes for each.go build ./coderd/...andgo test ./coderd/x/chatd/chatopenai/ ./coderd/x/chatd/chatprovider/pass against the new pin.