quantumnous/new-api is a self-hosted AI gateway that connects the APIs of multiple AI model providers on one server and exposes them to clients in a consistent form. It is written in Go and licensed under AGPL-3.0. The latest public release is v1.0.0-rc.45, marked as a release candidate, and the provided record contains no changelog for this release.
The project is a unified AI model hub that bundles multiple models. Its core function is converting different LLMs into OpenAI-compatible, Claude-compatible, and Gemini-compatible formats for delivery. The repository was created on November 10, 2023.
The README describes the project as a self-hosted AI gateway for applications, agents, and teams. It connects upstream model services — that is, the APIs of external model providers — to provide clients with a consistent API, and manages routing, access permissions, usage, and cost from a web console. Multiple providers can be attached together to operate a private multi-model service. The GitHub stargazer count is 50,341.
However, the provided information contains no performance figures such as throughput or response time. In the evaluation scores, only the momentum item is filled in at 7.5 out of 10, while the novelty, technology, and quality items are empty. It is difficult to judge the scale and stability in real-world operation from this information alone. If performance is a premise, the release notes for v1.0.0-rc.45 must be checked directly.