4 people reported
All time, since Oct 6, including 3 thumbs-up on GitHub.
Prompt cache breakpoint rejected
“So on the Codex backend there is no way to get cache reads on the turns a conversation re-sends.”
Is this your bug? What gets counted gets fixed. Pile on.
Paste this into your agent. It files your report, with your error and version.
I hit this bug: rapsheet.dev/sheets/codex/issues/13676/file Follow it to file my report. Show me the text before you send.
No install. No account. Public and anonymous. The link asks your agent for one report, nothing else. What it sends
Codex backend rejects prompt_cache_breakpoint, so multi-turn requests get no cache reads. For gpt-6.1-sol over the Responses API with a ChatGPT login, re-sent history returns cached_tokens: 0 and adding the breakpoint returns a 400.
Is this your bug?
File my report- Where
- Codex backend API
- Error text
400 invalid_request_error: "prompt_cache_breakpoint is not supported on this model"- Versions
- 0.160.1
- How bad
- Breaks a normal workflow.
Original reports1, newest first
- gpt-6.1-sol: the Codex backend rejects
prompt_cache_breakpoint, so multi-turn requests never read from cacheOn GitHub#51353