AutoGen vs CrewAI: Which Is More Actively Maintained? Live GitHub Data
AutoGen is Microsoft’s open-source framework for building multi-agent applications, while CrewAI is a Python framework for orchestrating teams of role-based agents. This page compares how actively each project is maintained, using live data from GitHub.
GitHub data checked 6 minutes ago.
Side by side
| Measure | AutoGen | CrewAI |
|---|---|---|
| Status | Archived | Active |
| Commits, last 91 days | 1 | 328 |
| Same 91 days a year earlier | 66 | 198 |
| Change vs last year | −98% | +66% |
| Last commit | 16 Oct 2026 | 18 Oct 2026 |
| GitHub stars | 61,465 | 59,837 |
| Stars gained, daily snapshots | +26 in 2 days | +61 in 2 days |
| Licence | CC-BY-4.0 | MIT |
| Managed cloud | Not checked | Not checked |
| Visitor rating | —0 of 3 ratings needed | —0 of 3 ratings needed |
What the numbers say
In the last 91 days, AutoGen’s default branch received 1 commit and CrewAI’s received 328. That is about 328 times as many for CrewAI, though projects that squash each pull request into one commit show fewer commits for the same work, so treat the gap as a rough signal. Against the same 91 days a year earlier, AutoGen is at −98% and CrewAI at +66%. The index labels AutoGen “archived” and CrewAI “active”. AutoGen has 61,465 GitHub stars and CrewAI has 59,837; stars show how many people bookmarked a project, not how many use it. Their licences differ: AutoGen uses CC-BY-4.0 and CrewAI uses MIT.
What this comparison does not tell you
These are maintenance signals only. They do not measure speed, output quality, features, documentation or how easy either tool is to run, and we have not benchmarked either tool for this page. A quiet repository can still be a good fit for a stable job, and a busy one can still break between releases. Use the numbers to decide what to check next, not which tool is better.
Each tool’s full record
Related comparisons
How this is measured: commits on each repository’s default branch counted through the GitHub API, the last 91 days against the same 91 days a year earlier; stars, licence and archived flag from the repository; README checked daily for maintenance notices. Visitor ratings come from visitors to each tool’s page, not from EvalAITools.