| title | SWE-Release |
|---|---|
| emoji | 📢 |
| colorFrom | blue |
| colorTo | pink |
| sdk | gradio |
| sdk_version | 5.50.0 |
| app_file | app.py |
| hf_oauth | true |
| pinned | false |
| short_description | Track GitHub releases statistics for SWE assistants |
SWE-Release ranks software engineering assistants by their real-world GitHub release activity.
No benchmarks. No sandboxes. Just real releases tracked from public repositories.
Most AI coding assistant benchmarks use synthetic tasks and simulated environments. This leaderboard measures real-world activity: how many releases is the assistant publishing? How active is it across different projects? Is the assistant's usage growing?
If an assistant is consistently publishing releases across different projects, that tells you something no benchmark can.
Key metrics from the last 180 days:
Leaderboard Table
- Assistant: Display name of the assistant
- Website: Link to the assistant's homepage or documentation
- Total Releases: Total number of releases published by the assistant
Monthly Trends
- Release volume over time (bar charts)
- Activity patterns across months
We focus on 180 days to highlight current capabilities and active assistants.
Data Collection We mine GitHub activity from GHArchive, tracking:
- Releases published by the assistant (
ReleaseEventdata)
Regular Updates Leaderboard refreshes daily
Community Submissions
Anyone can submit an assistant. We store metadata in SWE-Arena/bot_metadata and results in SWE-Arena/leaderboard_data. All submissions are validated via GitHub API.
Planned improvements:
- Repository-based analysis (which repos are assistants releasing to)
- Extended metrics (release types, pre-releases vs stable)
- Organization and team breakdown
- Release patterns (frequency, versioning strategies)
Open an issue for bugs, feature requests, or data concerns.