Expand description
ChatGPT Codex provider foundations.
This provider targets the ChatGPT-backed Codex Responses endpoint. The
endpoint is not a stable OpenAI Platform API, so the public model ids keep a
separate chatgpt-codex/ prefix and status copy labels it experimental.
Structs§
- Chat
GptCodex Client - Concrete ChatGPT Codex API client.
Enums§
- Responses
SseEvent - Parsed ChatGPT Codex Responses stream event.
Constants§
- BASE_
URL - ChatGPT Codex backend base URL.
- DEFAULT_
RECOMMENDED_ MAX_ TOKENS - Recommended request budget for known ChatGPT Codex models.
- DISPLAY_
MODEL_ PREFIX - Display-only prefix for ChatGPT Codex models in the terminal UI.
- MODEL_
PREFIX - User/config-facing model id prefix.
- RESPONSES_
URL - ChatGPT Codex Responses endpoint.
Functions§
- display_
model_ id - Render a concise display label without changing the configured model id.
- error_
message - Convert a ChatGPT Codex error into a human-readable failure string.
- is_
model_ id - Whether
modelis a ChatGPT Codex model id. - is_
retryable_ error - known_
models - Current ChatGPT Codex models from the provider expansion plan.
- parse_
responses_ sse_ chunk - Parse raw ChatGPT Codex Responses SSE data lines.
- parse_
responses_ sse_ event - Parse one ChatGPT Codex Responses SSE
data:payload. - probe_
env_ or_ dotenv_ authentication - Verify the configured ChatGPT Codex credential using the model catalog.
- raw_
model_ id - Strip
chatgpt-codex/from a model id. - reasoning_
options - Return the currently verified ChatGPT Codex reasoning controls.
- record_
response_ items - Extend an in-memory Responses history after one streamed response.
- record_
tool_ output - Append a tool result to an in-memory Responses history.
- supports_
reasoning_ effort - Whether this ChatGPT Codex model supports GPT-5.6 reasoning controls.