Responses API

Not yet available. The Responses API is on the roadmap. The page below documents the intended behavior.

The Responses API (POST /v1/responses) is a stateful alternative to chat completions. Instead of sending the full conversation history on every request, you create a response once and the server manages the thread — letting you append new turns cheaply without re-uploading prior context.

Overview#

Today the OpenDunes API exposes a single inference endpoint, POST /v1/chat/completions, which is stateless — your client holds the message history. The Responses API adds a complementary stateful surface that tracks conversation state on the server, reducing payload size and simplifying multi-turn agents.

Sub-pages#

• Basic Usage — create and continue a response thread.
• Reasoning — enable extended thinking on supported models.
• Tool Calling — register functions the model can invoke.
• Web Search — let the model retrieve live information.
• Errors — error codes specific to the Responses API.

Relationship to Chat Completions#

The Responses API is additive. Everything that works with POST /v1/chat/completions — model slugs, DA billing, streaming, API keys — works the same way here. The difference is lifecycle management: responses carry an id you reference in subsequent calls instead of resending the full messages array.

Pricing#

Billing follows the same per-token DA model as chat completions. Cached tokens from a stored conversation context will be billed at a reduced rate once Prompt Caching is live.