About
Access over 1000 models across 85+ labs, covering text, image, video, voice, audio, and embeddings. You can simplify your infrastructure by using one API to reach the entire landscape of AI models. What makes it different from a simple proxy: Drop-in OpenAI SDK โ change base_url, keep everything else. Works from Python, Node, Go, Rust, anywhere the OpenAI SDK exists. Aggregator-only primitives โ sub-accounts for end customers, per-feature cost attribution via x-aig-tag, replay any past request on a different model, eval-driven routing aliases that auto-bind to your winning model. Agent-first DX โ point Claude Code or Cursor at /llms.txt and it autoconfigures a full multi-modal pipeline on the first try. Models live today (1000+ total): Claude Opus 4.7, GPT-5.5, Gemini 3.1 Pro, Kimi K2.6, Seedance 2, Happy Horse, FLUX 2 Klein, Imagen 4, Nano-Banana, Veo 3.1, Seedream 5, Whisper, Nova-3, Aura 2, ElevenLabs, Inworld, BGE, and more. Try it: https://aigateway.sh