Skip to main content
The Impossibl AI API is a single API for all AI providers. It routes to OpenAI, Anthropic, Google, xAI, Z.ai, and Azure Foundry-hosted models (Llama, Mistral, DeepSeek, …), and bills you from a prepaid credit balance. It speaks the providers’ own wire formats so existing tools work unchanged:
  • POST /v1/chat/completions — OpenAI Chat Completions (drop-in for any OpenAI SDK)
  • POST /v1/responses — OpenAI Responses API
  • POST /v1/messages — Anthropic Messages API (drop-in for the anthropic SDK)
  • POST /v1beta/models/{model}:generateContent — Gemini-native generateContent
  • POST /v1/audio/transcriptions — completed-file speech-to-text
  • POST /v1/images/generations — OpenAI Images (image output)
Pick a model with the provider/model convention, e.g. anthropic/claude-sonnet-4-5. See GET /v1/models for the full list and pricing. Beyond text, every surface accepts image input, Gemini models accept audio and video input, and every surface supports tool calling in relay mode. Chat output is text; images are generated through openai/gpt-image-2 and completed audio files are transcribed with openai/gpt-transcribe.

​
Agent-first

Everything is built so an agent can self-serve with no human in the loop: There’s no dashboard yet — humans sign up the same way, with the two requests in the quickstart, straight from a terminal.

Agent quickstart

Register a funded account and make your first call in two requests.