Responses API
OpenAI-Compatible Responses API
Responses API is useful for newer agent-style clients and models that support the OpenAI Responses wire format. Confirm support before choosing it for a tool.
Who this is for
Developers configuring newer clients such as modern Codex flows.
Configuration reference
Values to confirm before setup
Responses endpoint family
Check the official Responses API compatibility page
Codex wire API
responses for supported models; chat for older chat-completion-only models
Setup flow
Practical steps
- 01Confirm the selected model supports Responses API.
- 02Use the documented Responses-compatible endpoint.
- 03Configure the client wire protocol as `responses` if the tool requires an explicit setting.
- 04Test a single prompt, then test conversation continuation if needed.
- 05Fall back to Chat Completions if the model or tool does not support Responses.
When to use it
Use Responses API when the client or model specifically benefits from it, such as newer agent workflows. Do not enable it just because the name sounds newer.
Quote note
The written quote should name the API protocol because tool support, model selection, and troubleshooting depend on it.
Common mistakes
Check these before escalating
- Responses-compatible endpoint paths can differ from Chat Completions paths.
- A tool configured for `responses` may fail against a chat-only model.
- The model list changes, so verify before delivery.
Related guides
Codex
Configure Codex with Model Studio or Token Plan credentials, then verify the selected model before handing it to developers.
OpenAI-Compatible Chat API
Most OpenAI-compatible integrations need only three changes: API key, base URL, and model name. The hard part is choosing the correct plan and endpoint.
Choose a Model
Select the model by workload, context length, modality, region, latency, and cost. The model list changes over time, so final quotes should reference the official current catalog.