// the find
labring/aiproxy
AI Proxy is a high performance AI gateway using OpenAI / Claude / Gemini protocol as the entry point. It features intelligent error handling, multi-channel management, and comprehensive monitoring. With support for multiple models, rate limiting, and multi-tenant isolation.
An AI gateway written in Go that proxies requests across OpenAI, Anthropic, and Gemini providers with protocol conversion, multi-tenant isolation, and per-group rate limiting. Built primarily to power the Sealos platform (also by labring), but usable standalone. Targets teams reselling or internally distributing AI API access.
The plugin system solves real problems rather than invented ones: the stream-fake plugin keeps non-streaming requests alive by internally converting them to streaming, which actually works around timeout issues that bite real deployments. The think-split plugin that extracts <think> tags into reasoning_content fields saves every consumer from reinventing that parsing. Protocol conversion is genuinely bidirectional — Chat/Claude/Gemini requests can hit responses-API-only models transparently. The built-in tokenizer avoids the tiktoken Python dependency mess that plagues other Go gateways.
The project is a labring internal tool that got open-sourced, and it shows — Sealos deployment is a first-class path while the standalone story is thinner. The README's Claude API example has `"model": "gpt-5"` in the request body, which suggests the docs aren't validated against real runs. Monitoring is a management panel dashboard rather than Prometheus metrics or OpenTelemetry traces, so you can't plug this into an existing observability stack without extra work. Billing configuration is opaque: enabling it and configuring per-group pricing involves enough database-level state that a misconfiguration could silently under- or over-charge groups with no obvious audit path.