Skip to main content

Private Model Hub

(Experimental Feature — enable in Profile > Settings > Experimental Features)

The Private Model Hub gives you access to an exclusive collection of privately hosted cutting-edge AI models, including DeepSeek, Qwen, and other frontier models that may not be available through mainstream providers.

Availability: This feature may be enabled or disabled at the organizational level by your administrator. If the toggle is grayed out with "Disabled by administrator," contact your organization admin to request access.

What It Does​

When enabled, additional models appear in your model selector alongside standard provider models (OpenAI, Anthropic, Google, etc.). These models are hosted on private infrastructure and offer:

  • Frontier models — Access models from providers and research labs not yet integrated as mainstream providers
  • Alternative architectures — Try models with different strengths and specializations
  • Privacy-first — Models run on dedicated infrastructure

How to Enable​

  1. Click your avatar in the sidebar footer to open your Profile
  2. Go to the Settings tab
  3. Scroll to Experimental Features
  4. Toggle Private Model Hub on

Once enabled, the new models will appear in your model selector when creating or configuring a notebook.

Selecting Private Models​

Open the model selector in any notebook and look for the privately hosted models in the list. They work just like any other model — select one, adjust temperature and other settings as needed, and start chatting.