Edgee is an AI gateway that optimizes prompts at the edge using intelligent token compression, removing redundancy while preserving meaning. The compressed requests are then forwarded to your LLM provider of choice. It serves as an edge intelligence layer for AI traffic behind a single OpenAI-compatible API.

Key features include token compression that reduces prompt size without losing intent to lower costs and latency, especially for long contexts, RAG pipelines, and multi-turn agents. The platform offers universal compatibility with any LLM provider including OpenAI, Anthropic, Gemini, xAI, and Mistral. Cost governance allows tagging requests with custom metadata to track usage and costs, with alerts for spending spikes. Additional capabilities include edge tools for invoking shared tools or deploying private tools, observability for monitoring latency and errors, edge models for running small fast models at the edge, and private models for deploying serverless open-source LLMs.

Edgee works by sitting between your application and LLM providers. Your application calls Edgee, which applies policies at the edge including routing, privacy controls, and retries, then forwards the request to the best provider for the job. The system normalizes responses across models so you can switch providers easily, allows observing and debugging production AI traffic end-to-end, and provides cost control with routing policies and caching.

The primary benefit is reducing LLM costs by intelligently compressing prompts at the edge, with up to 50% input token reduction. Use cases include handling long contexts, RAG pipelines, multi-turn agents, classifying and routing requests, and deploying private models alongside public providers.

Edgee targets developers and organizations building AI applications who need to optimize costs while maintaining performance. It integrates with major LLM providers and supports programming languages including TypeScript, Python, Go, and Rust through an OpenAI-compatible API.