[Additive] OpenAI: 4 new response fields

OpenAI's API changelog lists 4 new response fields, all variants of input_tokens_details.cache_write_tokens_details across ResponseUsage, BetaResponseUsage, BetaResponse.usage and Response.usage, alongside 2 breaking changes: prompt_cache_options is no longer accepted in CreateModelResponseProperties and BetaCreateModelResponseProperties and is now silently ignored rather than rejected, and a value can now return "max_output_tokens", which breaks exhaustive switches, strict parsers and generated client enums. The silent-ignore change matters most because affected calls still succeed while behaving differently.

OpenAI https://driftwatch.et/apis/openai 2 breaking4 additive - Breakingdirection flip: their tools call this safe prompt cache options is no longer accepted https://driftwatch.et/changes/openai-2026-11-03-7094f95d Usually ignored rather than rejected, so the call succeeds and does something subtly different. 2 places: CreateModelResponseProperties.prompt cache options, BetaCreateModelResponseProperties.prompt cache options Code that breaks await client.post "/…", { // prompt cache options is now ignored, not rejected prompt cache options: value, } ; - Breakingdirection flip: their tools call this safe value can now return "max output tokens" https://driftwatch.et/changes/openai-2026-11-03-2cc3309f Producer-side tools call this additive. It breaks exhaustive switches, strict parsers and generated client enums. SessionTurnErrorCodeResource · reachable from 6 operations Code that breaks switch response { // existing branches … // no branch for "max output tokens" default: throw new Error "unexpected value" ; } 4 new response fields - ResponseUsage.input tokens details.cache write tokens details - BetaResponseUsage.input tokens details.cache write tokens details - BetaResponse.usage.input tokens details.cache write tokens details - Response.usage.input tokens details.cache write tokens details