{"repo":"elixir-vibe/llm_proxy","free":true,"listed":false,"github":"https://github.com/elixir-vibe/llm_proxy","clone":"git clone https://github.com/elixir-vibe/llm_proxy.git","description":"Self-hosted, Elixir-native alternative to LiteLLM: one API for multi-provider routing, fallbacks, budgets, keys, and observability.","language":"Elixir","stars":21,"topics":["ai","anthropic","elixir","gateway","llm","observability","openai","phoenix","proxy","quotas"],"license":"MIT","category":"ai-agents","readme_excerpt":"LLMProxy A self-hosted, Elixir-native alternative to LiteLLM. Put one API in front of OpenAI, Anthropic, OpenRouter, Kimi Code, OpenAI Codex, and custom OpenAI-compatible providers—then switch models, add fallbacks, and control spend without changing every application that calls them. LLMProxy gives applications stable model names such as fast , smart , or cheap . You decide where those names run, how traffic fails over, who may use them, and how much they may spend. Provider credentials and usage data stay in infrastructure you control. Use LLMProxy directly inside an Elixir application: Or run it as an OpenAI-compatible service: Why LLMProxy - Connect your applications once. Point any OpenAI-compatible client at LLMProxy, use the Anthropic Messages API, or call it directly from Elixir. Stop duplicating provider adapters, authentication, retries, and usage tracking in every product. - Change providers without changing clients. Applications request a public name such as fast ; you can move it from OpenAI to Anthropic, OpenRouter, or a private endpoint by changing gateway configuration instead of application code. - Keep working when a provider does not. Route across multiple models, providers, regions, or credentials. Timeouts, retries, health-aware fallbacks, circuit breakers, and load-balancing strategies keep individual failures away from your users. - Control spend before the invoice arrives. Issue keys per application, customer, or team; restrict which models they can ca","default_branch":null,"files":null,"tree":[],"storefront":"/r/elixir-vibe","claimed":false,"request_supported":{"post":"https://gitbuyer.com/r/elixir-vibe/llm_proxy/request-supported","requests":0},"note":"indexed from public GitHub; nothing is for sale on this page. Clone it from GitHub. Paid listings live at /search."}