FastChat

FastChat offers an open platform for training, serving, and evaluating large language models, and hosts release repositories for Vicuna and Chatbot Arena.

github.com/lm-sys/FastChat
FastChat homepage screenshot

About FastChat

FastChat is an open-source platform for training, serving and evaluating large language model chatbots. It provides the training and evaluation code behind models such as Vicuna and MT-Bench, plus a distributed multi-model serving system with a web UI and OpenAI-compatible RESTful APIs. The project also powers Chatbot Arena (lmarena.ai), which has served over 10 million chat requests for 70+ LLMs and collected over 1.5 million human votes for an online LLM Elo leaderboard.

It can be installed with pip or built from source and run from the command line, with model weights for Vicuna, LongChat and FastChat-T5 downloaded automatically from Hugging Face repositories into a local cache folder. Vicuna is based on Llama 2 and is intended to be used under Llama's model license. Some context lengths and CLI output modes carry extra requirements: 16K Vicuna versions need transformers >= 4.31, and the rich-text CLI style is described as experimental and may not work on certain terminals. No pricing information is published for the repository.

Key features

  • Training and evaluation code for models such as Vicuna and MT-Bench
  • Distributed multi-model serving system with web UI
  • OpenAI-compatible RESTful APIs for served models
  • Command-line inference with automatic Hugging Face weight downloads
  • Installation via pip or from source, with Docker files included
  • Supports many model families including Llama 2, Falcon, and ChatGLM
Best for: Developers and researchers who need open-source tooling to train, serve and evaluate large language model chatbots.
Good for
  • Open-source platform covering training, serving and evaluation of LLM chatbots
  • Underlies Chatbot Arena, which has served over 10 million chat requests for 70+ LLMs
  • Supports a wide range of model families and provides OpenAI-compatible APIs
Watch out for
  • Vicuna weights are based on Llama 2 and should be used under Llama's model license
  • 16K context versions require transformers >= 4.31
  • Rich-text CLI output is experimental and may not work on some terminals

FastChat at a glance

Type
Open-source project
Pricing
Not confirmed. Check the site
Open source
Yes · Source code
Language
English
Last checked
Oct 5, 2026 · Online when checked
Listed
Oct 5, 2026

FastChat alternatives

Tools and sites you could use instead, ranked by how closely they match.

All 24 alternatives

Popular in AI Infrastructure & MLOps

Well-known listings from the same topic.

All 341

FastChat FAQ

What is FastChat used for?
It is an open platform for training, serving and evaluating large language model based chatbots, and includes the training and evaluation code for models such as Vicuna and MT-Bench.
Do I need to download model weights separately?
The provided commands start chatting and automatically download weights from Hugging Face repositories, which are stored in a cache folder in the user's home directory.
What license applies to Vicuna?
Vicuna is based on Llama 2 and should be used under Llama's model license.
Is something wrong or out of date?

Tell us and an editor will check it.

Own this website?

Claim the listing and suggest updates yourself. An editor checks every claim first.

Claim this listing