Models Prices Dashboard

Get started

Overview Quickstart Authentication and API keys Models and pricing

Text

Chat completions Streaming Tool calling Structured output Fallbacks

Media

How media jobs work Images Video Speech Transcription Music Webhooks Inputs

Routing and data

Provider preferences Data retention Idempotency Request ids and metadata

Account

Credits and billing Rate limits Key restrictions and spend caps

Reference

Errors Changelog

Overview Cancel a job POST Create a chat completion POST Generate an image POST Generate music POST Generate speech POST Transcribe audio POST Generate a video POST Get a job GET Get a model GET List models GET

Integrations

OpenAI SDK LangChain Vercel AI SDK Any OpenAI-compatible client Not supported

API reference

Overview

Every endpoint, its request, its response and code samples.

Every endpoint is served from https://api.opinference.com and takes your key as Authorization: Bearer <key> (see Authentication). Pick an endpoint from the sidebar. Errors share one envelope, described in Errors.

Changelog

What changed and when.

Cancel a job POST

Ask for a queued or processing job to stop. A job that has already finished answers `409`.