Martian
Martian operates an LLM model router and gateway that dynamically routes each request to the best underlying model across providers for the optimal balance of quality, latency, and cost. The Martian Gateway exposes a drop-in, OpenAI-compatible REST API (and an Anthropic Messages-compatible surface) so applications can route across a large catalog of models by changing only the base URL.
APIs
Martian Chat Completions API
OpenAI-compatible chat completions routed dynamically across a large catalog of provider models. Models are addressed with a provider/model-name string, with optional routing co...
Martian Messages API
Anthropic Messages-compatible endpoint that accepts the Anthropic Messages request shape and routes it through the Martian Gateway, enabling Anthropic SDKs and frameworks to use...
Martian Models API
Lists every model currently supported by the Martian Gateway, returning the provider/model-name identifiers usable as routing targets in chat completions and messages requests.
Event Specifications
Martian Gateway Chat Completions Streaming (HTTP + SSE)
AsyncAPI 2.6 description of the Martian Gateway's **chat completion streaming** surface. Martian does not publish a WebSocket API. The Martian Gateway is an OpenAI-compatible mo...
ASYNCAPI