TurboLLM

TurboLLM is a local runner for open language models that works on the user's own computer with no cloud, no account and no internet connection needed. It starts with a single npx turbollm command or a standalone app for Windows, macOS, Linux and Android, and serves a browser interface on localhost that phones, tablets and laptops on the same network can also use. It is aimed at people who want to run GGUF models on their own hardware and keep prompts, chats and files on their machine. It works with any llama.cpp compatible engine, including custom builds and community forks, and covers llama.cpp, ik_llama.cpp, TurboQuant, KoboldCpp, llamafile, MLX, vLLM and SGLang. On load it benchmarks the hardware, derives fast defaults and shows a video memory fit verdict before a model is loaded. Speed figures in the model list are tokens per second measured on the user's own machine during real generation. It exposes OpenAI compatible and Anthropic compatible APIs, so the Claude Code command line tool and other existing tools can be pointed at a local model with one command. A local coding agent, built on the pi SDK, can read, edit, run commands and report diffs in a repository folder on the user's own GPU. iOS is not supported. The site's structured data lists a price of zero dollars and no paid plans are published.