New API is an open-source AI model gateway with 50,300+ stars on GitHub that aggregates many LLM providers behind one managed hub and converts between OpenAI-, Claude- and Gemini-compatible API formats.
Anyone juggling API keys for OpenAI, Claude, Gemini, DeepSeek and a dozen other providers ends up needing one place to route, meter and control that usage. New API, developed by the QuantumNous community in the lineage of the one-api project, is exactly that hub, self-hosted under your control. You register provider channels once, issue your own access tokens, and call every model through a single unified endpoint. Its signature capability is format cross-conversion: an application written against the OpenAI API can talk to Claude or Gemini models, and vice versa, without code changes. Around the gateway sit the operations features teams need — per-token and per-user quotas, usage and billing statistics, channel load balancing with automatic failover, and a web admin console for managing users, models and pricing ratios. Deployment is a single Docker command, and the project is updated almost daily.
New API is free and open source under the AGPL-3.0 license. You host it yourself and only pay the underlying model providers for the usage you route through it — the gateway itself charges nothing.
Developers, teams and enterprises that manage many model API keys and want a single controlled, metered entry point — for internal platforms, shared team access or reselling-style distribution with proper accounting.