New API

Overview

New API is an open source AI gateway for applications, agents and teams. You connect upstream model services as channels, such as OpenAI, Anthropic, Google Gemini, Azure OpenAI, AWS Bedrock, DeepSeek or any OpenAI compatible endpoint, and New API exposes them to your clients through one consistent API. It routes each request to a channel that serves the requested model, converts between provider formats where needed, and streams the response back. A web console manages channels, models and their prices, users and groups, per client API keys with quotas and expiry, and records every call with its tokens, cost and latency so you can see who used what. Data lives in an embedded SQLite database, so a single virtual machine runs the whole gateway. It suits teams who want one audited entry point to many model providers, and developers who want to switch providers without reconfiguring every client.

Common uses

• Give a team one OpenAI compatible endpoint in front of several model providers
• Issue per application API keys with quotas and expiry, and see usage and cost per key
• Switch or fail over between providers without changing client configuration

See it running

Real screenshots taken while testing this image against its deployment guide.