High-performance LLM reverse proxy and upstream gateway. Compatible with OpenAI and Ollama protocols with SubLLM policy routing, sequential failover, and local mesh bridging.
Point any standard SDK, LangChain, or agent framework at proxy.oqllm.com or your local daemon:
export OPENAI_API_BASE=http://127.0.0.1:11437/v1 export OPENAI_API_KEY=oqllm_live_... # Routes model "oql" for scenario generation, # while any standard model passes through to SubLLM
Compatible with native Ollama command-line tools and web UIs:
export OLLAMA_HOST=http://127.0.0.1:11437 ollama list ollama run qwen2.5-coder:7b
Deterministic sequential failover across model providers:
Primary: Z.AI (glm-5.3) Failover: OpenRouter → Codex → Gemini Fallback: Local Ollama (offline mode)
Federate remote worker machines, edge devices, and mini PCs:
oqllm-proxy serve --host 0.0.0.0 --port 11437 \ --mesh-uplink https://api.oqllm.com \ --api-key oqllm_live_YOUR_KEY \ --node-id edge-node-01
Install the enterprise client SDK from the private GitHub release or repository:
pip install "oqllm-proxy @ git+https://github.com/oqllm/proxy.git"