Gemini Balance is an open-source proxy service that pools multiple Google Gemini API keys and load-balances requests across them. Built for developers who keep hitting per-key rate limits, it polls and rotates keys automatically, tracks the health and usage of each key, and exposes both Gemini-native and OpenAI-compatible endpoints. The project has earned more than 5,700 stars on GitHub.
You deploy Gemini Balance yourself, add your Gemini API keys, and point your applications at it as if it were the Gemini API. Requests are distributed across the key pool, failed or rate-limited keys are skipped, and usage statistics show how each key is performing. Because it also speaks the OpenAI API format, existing tools and clients built for OpenAI can talk to Gemini models, including Imagen image generation, with little more than a base-URL change. Configuration is deliberately simple: keys are managed through a web panel and configuration files, requests can be filtered and logged, and the service can sit in front of anything from personal scripts to shared team endpoints. For developers in regions or plans where a single Gemini key is heavily throttled, pooling several keys is often the difference between a demo that stalls and an application that stays responsive.
Free. Gemini Balance is an open-source, self-hosted project; the only cost is whatever you pay Google for the Gemini API keys you pool in it.
Developers and small teams who use the Gemini API heavily and want higher effective quotas, better reliability and one stable endpoint for all their applications.