fix: add Databricks models databricks-gpt-5 +2 more#650
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
github-actions
Bot
requested review from
Alex Z (CLowbrow),
aswink,
Caitlin Pinn (cpinn),
Erin McNulty (erin2722) and
Ken Jiang (knjiang)
![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: e8a4593463
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "Codex (@codex) review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "Codex (@codex) address that feedback".
| "flavor": "chat", | ||
| "multimodal": true, | ||
| "displayName": "GPT-5", | ||
| "max_input_tokens": 400000, |
There was a problem hiding this comment.
max_input_tokens is documented in this repo as an input-only limit, but these new Databricks GPT-5 entries set it to 400000 (the total context window) instead of the effective prompt cap used elsewhere for GPT-5-family models (272000). This overstates capacity by 128k tokens, so any caller that trusts model metadata can send prompts that pass local checks but are rejected upstream for token-limit overflow.
Useful? React with 👍 / 👎.
Resolve conflicts: keep main + databricks-gpt-5, databricks-gpt-5-1, databricks-gpt-5-mini (all return ENDPOINT_NOT_FOUND — not on CI plan). FIX (codex-bot P2): correct max_input_tokens 400000 -> 272000 (GPT-5 family input cap; 272000 in + 128000 out = 400000 total) on the 3 new models AND on databricks-gpt-5-nano (same bug, merged earlier via #652). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Re-resolve after main advanced: keep new main + databricks-gpt-5/-5-1/-5-mini (max_input 272000) and the databricks-gpt-5-nano max_input 400000->272000 fix. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Resolve conflicts: keep main + databricks-gpt-5-4, databricks-gpt-5-4-nano, databricks-gpt-5-2 (all ENDPOINT_NOT_FOUND — not on CI plan). FIX: max_input_tokens 400000 -> 272000 (same as #650; the databricks GPT-5 family uses a uniform 400000 total = 272000 input + 128000 output, so 400000 overstated the input cap). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Resolve conflicts: keep main + databricks-claude-opus-4-7, databricks-claude-sonnet-4-6, databricks-gpt-5-5. All ENDPOINT_NOT_FOUND (not on CI plan). FIX: databricks-gpt-5-5 max_input_tokens 400000 -> 272000 (databricks GPT-5 family input cap; same as #650/#647). Claude entries (1M) kept. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2 participants
fix: add Databricks models databricks-gpt-5 +2 more
Closes #641
Source issue: #641
Summary
databricks-gpt-5databricks-gpt-5-1databricks-gpt-5-minidatabricks-gpt-5databricks-gpt-5-1databricks-gpt-5-mini2
Verified metadata
Verification notes
No LLM verification step ran — model metadata was already complete in the issue.
sync_models vs proposed update
sync_models cross-check found differences. Official provider verification was used for the applied values, and sync_models discrepancies are listed below for review.