Open-source desktop app to run, serve and fine-tune text, image, video and audio models entirely on your own machine.
LocalAI
What is LocalAI?
Open-source, self-hosted AI engine that runs open models for chat, images, speech, video and music on your own CPU or GPU behind an OpenAI- and Anthropic-compatible API.
- Pricing
- Free
- Free plan
- Yes
- Category
- Coding & Development
- Made by
- LocalAI (Ettore Di Giacinto)
- Updated
LocalAI screenshots
LocalAI features & details
LocalAI is a self-hosted alternative to the OpenAI API, maintained in the open since March 2023 under the MIT licence. It runs open models on your own hardware — a plain CPU works, and NVIDIA, AMD, Intel and Apple GPUs make it faster — and exposes them through an API that speaks the OpenAI format, plus the Anthropic and ElevenLabs formats. Existing client code typically only needs its base URL changed to http://localhost:8080/v1.
It covers more than chat. The same server handles vision and tool calls, image, video and music generation, speech-to-text, text-to-speech, embeddings and reranking. Models come from a built-in gallery, and LocalAI picks and downloads a suitable backend for your machine from more than 80 supported engines, from llama.cpp to vLLM, 18 of them written by the project itself. It installs with a one-line script on Linux and macOS, or runs in a container, on Kubernetes, or in a distributed mode that spreads models across several machines.
On top of the API, recent versions add a web workspace for teams: per-user logins and chats, shared folders and projects with notes, Knowledge collections searched by meaning, saved agents, and an MCP client and server. Add-ons such as Office (generate Word, Excel, PowerPoint or PDF files from a request) and a network-isolated Python Sandbox are installed by an admin. Everything stays on your own server, and the software is free.
No reviews yet — sign in to write the first one
Apps similar to LocalAI
The closest matches are Unsloth and NobodyWho. See every LocalAI alternative compared by price and free plan.
Open-source inference engine that runs LLMs, vision and speech models fully on-device — no API keys, no cloud calls, from Swift to Godot.
8-29 MB automation models that run on phones, wearables and microcontrollers — tool calling, structured extraction and embeddings, fully offline.
Open-source local inference server that profiles your hardware, picks models that fit, and points your coding agent at them — free, private and offline.
Menu-bar LLM inference server for Apple Silicon, with continuous batching and tiered KV caching that keeps local models fast enough for real coding work.