Skip to main content
PoYo serves LLMs from several providers through four API formats. Each request sends a model id in model (or in the URL path for Gemini Native). A model only accepts the formats marked in the table below.

​
API formats

  • Chat Completions — POST /v1/chat/completions, OpenAI Chat Completions request shape
  • Responses — POST /v1/responses, OpenAI Responses request shape
  • Anthropic Messages — POST /v1/messages, Anthropic Messages request shape
  • Gemini Native — POST /v1beta/models/{model}:{method}, Google generateContent request shape

​
Models

All models support streaming through the stream option of each format, or streamGenerateContent for Gemini Native.

​
Notes

  • Only send a model to the formats marked for it. For example, gpt-5-6-sol only works with Responses. Other formats return HTTP 400 with the list of supported URIs.
  • Authenticate with Authorization: Bearer YOUR_API_KEY. Anthropic Messages also accepts x-api-key.
  • LLM responses are returned directly in each format’s native shape, without the code / data wrapper used by generation tasks. See each format page for the streaming event format.
  • LLM endpoints are currently not rate limited.