Self-Hosted AI with Hugging Face TGI
Text Generation Inference (TGI) is Hugging Face’s open-source server for running open models on your own NVIDIA GPUs, with an OpenAI-compatible API. Browse 63 open models that run on Hugging Face TGI, see how much GPU memory each one needs, and open any model for a step-by-step guide to chat with it in TypingMind.
Yi via Hugging Face TGI
01.AI · 2 sizes from 12.1 GB
Yi 1.5 is a high-performing, bilingual language model.
Zephyr-Beta via Hugging Face TGI
Hugging Face · 14.5 GB
Zephyr is a series of language models that are trained to act as helpful assistants.
Llama-2 via Hugging Face TGI
Meta · 3 sizes from 13.5 GB
Llama 2 is a collection of foundation language models ranging from 7B to 70B parameters.
TypingMind for Teams
Bring your self-hosted AI models to your team
Running Hugging Face TGI models on your own GPU server keeps your data private. To give your whole company the same private AI, run your models on a company server and connect them to TypingMind Teams. Everyone gets one AI workspace with your self-hosted models next to the cloud models you allow.
- Add your self-hosted models once for the whole team. Nobody has to install anything.
- Self-host TypingMind Teams in your own network, so chats never leave your infrastructure.
- Control who uses which model with SSO, roles, and user groups.
- Set usage limits and review every admin change in the audit log.
