sususu 883660fb81 fix(translator): deduct cache_write_tokens from input_tokens in openai/codex to claude (#5956) ...

Deduct cache_write_tokens and cache_creation_tokens from gross prompt/input tokens
when translating OpenAI Chat Completions and Codex Responses API to Anthropic Claude
Messages format to prevent downstream double-counting.

Include defensive guards against int64 overflow, clamp negative input_tokens to 0,
and support <= 0 fallback to cache_creation_tokens alias.

2026-09-19 13:31:17 +08:00

..

claude

fix(translator): deduct cache_write_tokens from input_tokens in openai/codex to claude (#5956)

2026-09-19 13:31:17 +08:00

gemini

fix(gemini): generate deterministic sequential IDs for Gemini tool call pairing in Codex/Claude conversions

2026-08-21 03:21:55 +08:00

interactions

perf(translator): batch assemble translation arrays before writing JSON

2026-08-16 01:25:28 +08:00

openai

fix(codex): normalize and support ultrafast service tier

2026-09-18 11:21:55 +08:00