AxonHub is an open-source AI gateway that lets you call more than 100 large language models through any SDK: it transparently translates requests between the OpenAI, Anthropic and Gemini API formats, so you can switch model providers without changing a line of client code. It has earned more than 5,300 stars on GitHub and is actively developed, with its latest updates in 2026.
AxonHub solves three everyday pains of building on LLMs: vendor lock-in, integration sprawl and missing observability. Point your existing OpenAI SDK code at AxonHub and it can route to Claude, GPT, Gemini, DeepSeek, Moonshot, Doubao, OpenRouter and many more providers, converting formats in both directions. Reliability features include smart load balancing across channels with automatic failover in under 100 milliseconds, while every request is traced end to end with per-request cost breakdowns across input, output and cache tokens, so budgets are visible in real time. Enterprise touches include role-based access control, per-key usage quotas and data isolation, plus an admin dashboard for managing channels, models and pricing. It deploys easily with Docker and supports text generation, image generation, reranking and embedding APIs.
AxonHub is free and open source (Apache-2.0/LGPL-3.0 components) for self-hosting. You pay only your model providers for usage.
AxonHub is for developers and teams who want one stable API in front of many model providers โ with failover, cost control and tracing โ instead of maintaining separate integrations for every vendor.