OpenAI (the reference)¶
The moving target. This page exists to document what OpenAI itself does, not because we expect anyone else to match it perfectly.
Surface¶
OpenAI implements the full canonical surface plus several extensions that haven't propagated to OSS servers:
/v1/responses— newer than chat completions, structured around responses (multi-turn, multi-output) instead of message arrays./v1/audio/speechstreaming./v1/threads,/v1/assistants,/v1/files— the Assistants API./v1/batches— async batch processing at half price./v1/realtime— WebSocket-based bidirectional streaming.
The catalog tracks Responses, speech, files, batches, uploads, and Realtime. Threads and Assistants remain intentionally out of scope: their multi-step stateful workflows do not fit the probe's two-phase endpoint contract.
Auth¶
Authorization: Bearer <key> is required. Pass --openai-api-key, or
set OPENAI_API_KEY (AIOC_API_KEY is the fallback).
OpenAI-Beta headers gate features that are in flight. The prober only
adds openai-beta: realtime=v1 for the Realtime WebSocket row.
Probing OpenAI itself¶
Why probe the reference? To detect spec drift in our own catalog.
If OpenAI changes the chat completions response shape and the prober
flags WARN against api.openai.com, the catalog is out of date,
not OpenAI.
Things to keep in mind¶
- Rate limits. Even with one shared discovery call and at most two requests per selected endpoint, a free-tier key may hit the per-minute cap on the chat probe. Use a paid tier or expect SKIP rows.
- Cost.
max_tokens=4keeps each chat call to ~¢0.001 ongpt-4o-mini. Image generation is ~¢4 per call (512x512, n=1ongpt-image-1). The full probe is roughly $0.05 against the cheapest tier. - Sandbox. OpenAI does not run in
sandbox=truemode by default. The probe sends real prompts; review the catalog's bodies if you're concerned about content.