How ranking works
AI-Sovereign tracks 7,188 AI projects from 2,862 organisations. Every number on the site comes from a public source, is refreshed every day, and links back to where it came from. Nobody can pay for a better rank.
What decides the rank
Each category is ranked by one public number, the way a market cap ranks coins:
- AI Models: downloads in the last 30 days on Hugging Face.
- AI Agents and Open-source AI: stars on GitHub.
- AI Apps: likes on Hugging Face Spaces.
- AI Research: community upvotes on Hugging Face Daily Papers.
- AI Model APIs: listed newest first, with input and output prices per million tokens and context size from OpenRouter. Sort by price or context to compare.
The score
The score from 0 to 100 shows where a project stands inside its own category: a score of 90 means it ranks above 90% of the projects we track there. It lets you compare a model with an agent at a glance without pretending downloads and stars are the same thing.
24h and 7d change
We keep a daily snapshot of each project's ranking number. The 24h and 7d columns show how much it grew since then, so fast risers stand out. New entries show a dash until they have history.
Companies
An organisation's strength is the sum of its projects' scores. A lab with many strong models, apps and repositories ranks above a team with one popular project.
Sources
- Hugging Face: models, Spaces and Daily Papers.
- GitHub: open-source AI projects and agents (projects with more than 500 stars on AI topics).
- OpenRouter: AI model APIs, prices and context sizes.
Data last refreshed 2026-11-22 06:44 UTC.