How to Get Involved or Contribute to Public AI Inference Utility

Updated

Public AI Inference Utility is a nonprofit, open-source service that makes public and sovereign AI models more accessible. There are two main ways to get involved: contributing code through pull requests, or applying to become a model or inference partner. The site's own call for contributors states plainly: "It's launched! Now we really need your help. Just send PRs – it's all open source."

Contributing as a Developer

The project is open source, and the primary contribution path is submitting pull requests. The site does not publish a separate contributor guide, repository link, or coding standards on the page, so the practical starting point is to treat the codebase itself as the source of truth.

What to expect

  • Input: your own code changes, fixes, or improvements to the open-source project.
  • Action: open a pull request against the project's repository.
  • Expected result: your contribution is reviewed and, if accepted, merged into the utility.

Common sticking points

  • The page does not list a repository URL, contribution guidelines, or a code of conduct. You will need to locate these from the project's public code hosting presence rather than the landing page.
  • There is no stated onboarding process, so expect to read existing code and issues before proposing changes.
  • Because the utility hosts models from multiple partners, changes that touch model serving or infrastructure may need coordination beyond a simple PR.

Becoming a Model or Inference Partner

The site explicitly invites organizations to join: "Interested in joining? Apply to become a model or inference partner." This is the route for institutions, labs, or compute providers that want their models or infrastructure hosted on the utility.

What the partnership involves

The page describes partners as "leading institutions and organizations" that provide "models and compute." Existing examples referenced on the site include:

Partner type Example from the site
Model Apertus, described as "the most powerful open-source language model ever released by a public institution"
Compute / deployment AWS Zurich and Intel, credited with helping "deploy public AI at national scale"
Hosted models EuroLLM, ALIA, and others mentioned as running on the layer built with those partners
Platform Hugging Face, where Public AI launched as "an official inference provider"

How to apply

  • Input: your organization's model or compute offering.
  • Action: use the "Apply to become a model or inference partner" link on the site.
  • Expected result: your model or inference capacity is evaluated for inclusion in the utility.

The page does not state eligibility criteria, timelines, or technical requirements for partners, so those details would come from the application process itself.

Other Ways to Engage

  • Read the mission: the About section explains that Public AI is "a nonprofit building products and organizing advocacy to support public AI model builders worldwide." Following that work is a lower-commitment way to stay involved.
  • Follow project news: the "Latest stories" section covers launches, partnerships, and advocacy pieces, which can point to where help is needed next.
  • Check the FAQ: the site's FAQ addresses common questions about the project's technical setup, roadmap, data handling, and privacy, which is useful context before contributing or partnering.

What the Site Does Not Specify

To avoid guessing: the page does not publish a repository link, a contributor license agreement, a partner pricing or cost model, or a stated review timeline for either pull requests or partnership applications. If any of those matter to your decision, confirm them through the project's public channels before committing time or resources.