An open, programmable decision layer for models and compute.
Fast LLM swapping with sleep/wake support, compatible with vllm, llama.cpp, etc. llama-swap fork.
High-performance Java LLM router & proxy with intelligent multi-model routing, failover, quota management, and OpenAI-compatible API endpoints.
Efficient Agent Router (EAR) is a Python-first orchestration service that selects and executes the best LLM for a request based on quality, cost, latency, context window, and safety constraints.
ftouter: drop-in fallback router for free and cheap-tier LLM APIs (Groq, OpenRouter, Mistral, Gemini). Automatically retries on the next model when one fails — no proxy server, no config. just pip install ftouter
An intelligent AI gateway built with FastAPI that classifies incoming prompts by complexity using TF-IDF and Logistic Regression, then routes each request to the cheapest capable LLM via Groq or Mistral. Includes semantic caching with ChromaDB to avoid repeated calls, PII redaction, and a background LLM-as-a-Judge system for automated quality audit
To associate your repository with the llmrouter topic, visit your repo's landing page and select "manage topics."