Commit message generation fails with “Request too large” error for large commits

[DefinitelyADev]

Description

When attempting to generate a commit message for a large commit, Fork fails with the following error:

OpenAI error: Request too large for gpt-4o in organization org-XXXXX on tokens per min (TPM):
Limit 30000, Requested 206842. The input or output tokens must be reduced in order to run successfully.
Visit https://platform.openai.com/account/rate-limits to learn more.

This prevents commit message generation entirely for commits with large diffs.

Steps to Reproduce

1. Stage a large number of files (e.g. lockfiles, snapshots, generated files, etc.)
2. Use AI → Generate Commit Message
3. Observe error dialog.

Expected Behavior

• Fork should be able to handle large diffs gracefully.
• Possible strategies:
  • Summarize only file paths (not full diffs).
  • Automatically ignore large/noisy files (e.g. package-lock.json, *.snap, *.min.js, etc.).
  • Option to choose which files/diffs to include in AI prompt.
  • Allow switching to a higher-throughput model (e.g. gpt-4o-mini) for commit messages.

Actual Behavior

• Fork sends the full diff to the model, exceeding OpenAI’s 30k TPM limit (my request hit ~206k).
• Error dialog appears and no commit message is generated.

Environment

• Fork version: 2.55.3
• OS: macOS 15.6 (24G84)
• OpenAI model: gpt-4o
• Organization TPM limit: 30000

Possible Solutions

• Add configurable max token budget per request.
• Add diff summarization (e.g. git diff --stat instead of full git diff).
• Provide a model selection option in Fork settings.
• Auto-ignore common large files from AI prompts.