OneRouter Docs

Build with OpenAI-compatible, Claude-native, and Gemini-native APIs through one controlled gateway.

What OneRouter is

OneRouter is a single gateway for multiple AI model protocols. It focuses on reliable access, clear request traces, accurate usage accounting, and a practical control layer for teams that run AI in production. The public API is designed to work with common SDKs while keeping model, billing, and observability information understandable.

Base URLs

Use the shortest protocol-specific base URL that matches your client. OpenAI-compatible SDKs use https://onerouter.com/v1. Claude native requests use https://onerouter.com with /v1/messages. Gemini native requests use https://onerouter.com/v1beta or the /gemini/* compatibility prefix.

ParameterTypeRequiredDescription
OpenAI-compatiblebase URLrequiredhttps://onerouter.com/v1
Claude nativebase URLrequiredhttps://onerouter.com
Gemini nativebase URLrequiredhttps://onerouter.com/v1beta

Authentication

Send your OneRouter API key with every request. OpenAI-compatible clients should use Authorization: Bearer YOUR_API_KEY. Gemini-native clients may also use x-goog-api-key. Keys are issued from the console and use the OneRouter key format.

Authorization: Bearer YOUR_API_KEY

Core API surfaces

ParameterTypeRequiredDescription
OpenAI-compatibleprotocolrequiredChat Completions, Responses, embeddings, images, audio, video, rerank, models.
Claude nativeprotocolrequiredMessages API and token counting routes for Anthropic-compatible clients.
Gemini nativeprotocolrequiredgenerateContent, streamGenerateContent, countTokens, and model listing through Gemini-compatible paths.
ConsoleaccountoptionalUsage logs, billing summaries, account balance, API key management, invoices, and support are available in the console rather than the public model API.

Model-dependent features

OneRouter keeps protocol compatibility and forwards model-specific parameters where supported. Features such as tool calling, structured output, reasoning, prompt caching, web search, image input, audio input, video generation, and provider-specific options depend on the selected model and upstream provider. Check the model detail page before relying on a feature in production.

Capability boundaries

Available parameters, response fields, task modes, and retention behavior can vary by model and provider. When a model returns a job ID, media URL, or provider-specific field, follow that model's documented response shape instead of assuming the same workflow applies to every model.