airllm
Run 70B+ LLMs on a single 4GB GPU — no quantization required.
Run OpenClaw AI agent with zero API cost, local LLM via AirLLM
