rootandClaude Fable 5 c97f52de82 gateway/v1: 402 affordability clamp-retry on the openrouter lane ...

Stripping max_tokens on a 402 made things worse (the precheck then
prices the model's own 65k maximum). Instead: parse the 'can only
afford N' budget from the error and retry once clamped to N when
N >= 256 — small answers survive a thin balance; anything less
surfaces the 402 verbatim, because the real fix is funding the
account the key belongs to.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

2026-08-28 17:12:53 -05:00

..

src

gateway/v1: 402 affordability clamp-retry on the openrouter lane

2026-08-28 17:12:53 -05:00

Cargo.toml

gateway: /v1/validate endpoint — Phase 43 v3 part 2

2026-04-27 07:40:27 -05:00