Smarter and Simpler AI Gateway

Prompts, embeddings, and models are now Provider Agnostic in LLM Proxy policies The problem with configuring policies per API spec Anthropic, OpenAI, and other model providers each have their own API format, and those formats keep changing. If you wanted to apply a policy to prompts, embeddings, or models, you had to configure it against the specific shape of that provider’s API. That meant the same policy, in concept, had to be rebuilt for every provider you supported.