OpenScores
OpenScores (openscores.io) is a project of the Open Intelligence Foundation that continuously scans open source AI projects on GitHub and evaluates each one across five dimensions — maturity, health, popularity, sustainability, and risk — using GitHub-derived metrics tracked over time. Every dimension is scored on a 0–100 scale, and the five are combined into a single overall White Score.
Why OpenScores matters
The open source AI ecosystem moves fast, and thousands of projects compete for attention. But stars and hype rarely tell you what you actually need to know: is this project professionally maintained? Is it still alive and responsive? Is it likely to keep going? Is it safe to build on? A project can be popular but unhealthy, mature but unknown, or healthy today but unsustainable — and judging by stars alone hides all of that. OpenScores answers these questions with data rather than guesswork, giving the community a transparent, consistent way to evaluate projects side by side. That makes it a critical tool for anyone adopting, funding, contributing to, or governing open source AI — a way to find the projects most worth investing in, and to spot those that may need help before they fail.
The White Score
The White Score is a five-part framework. Each dimension is independent and scored from 0 to 100 using GitHub metrics over rolling time windows, so a project’s standing on one axis never masks another:
- Maturity — structural readiness and governance; it answers “is the project professionally maintained?” Inputs include project age, release history and cadence, documentation, governance files (license, contributing, code of conduct, security policy), CI and tests, active maintainers, semantic versioning, and breaking-change signals.
- Health — current activity and responsiveness; it answers “is the project alive and functioning well?” Inputs include recent commit activity, how quickly issues and pull requests receive a first response, issue-close and PR-merge ratios, backlog pressure, contributor diversity, and CI success rate.
- Popularity — adoption and community traction; it answers “is it widely used or gaining attention?” Inputs include stars, forks and watchers, their recent growth, total contributors, and engagement from people outside the maintainer team.
- Sustainability — long-term viability; it answers “can the project keep going over time?” Inputs include the number of active maintainers, contributor diversity and retention, funding or sponsorship, organizational ownership, governance, release cadence, and maintainer workload.
- Risk — dependency and abandonment risk, where a higher score is worse; it answers “what could go wrong if we rely on it?” Inputs include the bus factor and contributor concentration, commit and release staleness, the absence of a security policy, stale issues, single-account ownership, and whether the project is archived.
How the overall score is composed
The leaderboard ranks projects by a weighted blend of all five dimensions:
Overall = 0.25 × Maturity + 0.25 × Health + 0.25 × Popularity + 0.15 × Sustainability + 0.10 × (100 − Risk)
The three core dimensions — maturity, health, and popularity — carry three quarters of the weight. Sustainability and risk enter at a reduced weight because several of their inputs (maintainer counts, commit and release recency, contributor concentration, governance) are already captured by maturity and health; counting them in full would double-weight the same signals. Risk is inverted so that higher is always better in the overall score, and an archived project has its risk pinned high — enough to subtract several points on its own. Every dimension is also shown and can be sorted on its own.
Leaderboard and catalog
OpenScores builds its catalog by sweeping GitHub across roughly 170 AI-related topics and keeping non-fork, non-archived repositories above a star threshold, then scoring each one. The leaderboard ranks the top projects — those carrying a recognized open source license — by their overall White Score, so you can see at a glance which open source AI projects lead the field. The full catalog lets you browse and compare every inventoried project. Whether you are choosing a dependency, scouting projects to support, or tracking the health of the ecosystem, it is an invaluable starting point.
Full details of every metric, weight, and benchmark are published in the OpenScores methodology.