Anvil

Tiny package, fastest ai/ml inference and training in the world Anvil - Forge anything. A zero-jank, single-binary local AI runtime. The definitive Ollama alternative for developers who refuse slow defaults, opaque protocols, and resource hogs. What is it? Anvil is a terminal-first local LLM tool that links llama.cpp directly natively. One binary. Zero runtime dependencies. No background daemon. No hidden blob storage. No telemetry. Just pure, raw, in-process inference.