Route requests to different LLM providers like OpenAI, Anthropic, or Gemini through one local endpoint.
Keep a long conversation pinned to the same model using session stickiness.
Track and optimize API spending across multiple providers and models.
Self host the community server for a team while keeping API keys local to each user.
| seasx/seasagi | 100/stock-analysis-markov | 100/tab-organizer | |
|---|---|---|---|
| Stars | 11 | 11 | 11 |
| Language | — | Java | JavaScript |
| Last pushed | — | 2016-12-25 | 2021-03-01 |
| Maintenance | — | Dormant | Dormant |
| Setup difficulty | moderate | easy | easy |
| Complexity | 3/5 | 3/5 | 1/5 |
| Audience | developer | researcher | general |
Figures from each repo's GitHub metadata at analysis time.
Client runs standalone with local API keys, the optional server component adds cloud sync and team features.
SeasAGI describes itself as a local LLM API gateway made up of three connected parts. The client is a desktop application built with Wails, Go, and React that runs a local gateway on a user's own machine so API keys never leave that computer. A community server adds an optional cloud control plane with basic accounts, channel management, and usage tracking, and a separate enterprise server adds paid features like billing, multi tenant management, and an admin dashboard. The README says the client works fully on its own without needing the server at all. On the client side, the README lists a local API endpoint that behaves like the OpenAI API, so it can be dropped into any tool that already talks to OpenAI by changing the base URL. It also describes routing strategies that fall back between providers or spread requests round robin, a session stickiness feature that keeps a single conversation on the same model for thirty minutes, and automatic handling of rate limit errors by temporarily deprioritizing a channel that keeps failing. Other listed features include a cost tracking and optimization view, support for eight different provider integrations such as OpenAI, Azure, Anthropic, Gemini, and Ollama, request logging and error charts, automatic health checks on API keys, and a built in chat window for testing setups directly. The community server, described as self hostable, covers basic login, channel setup, combo management, and usage statistics, while the enterprise edition is closed source and adds billing through Stripe along with team and governance features. Licensing is split by component. The client is GPL 3.0, the community server is AGPL 3.0, and the enterprise server is closed source, so anyone reusing or modifying the client or community server pieces must keep those parts open under their respective licenses, while the enterprise code is not open at all. The interface supports Chinese, English, Japanese, and Korean, and the README notes that API keys are stored in the operating system's keychain rather than in plain configuration files.
A local gateway app that unifies multiple LLM API providers behind one OpenAI compatible endpoint, with optional cloud and enterprise add ons.
The client is GPL v3.0 and the community server is AGPL v3.0, both requiring modified copies to stay open source, while the enterprise server is closed source.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.