Introducing Composer 2.5 (Cursor blog) May 18, 2026
Composer 2.5
Composer 2.5 places in the top three on CursorBench 3.1.
10
Benchmarks with a score
0
Ranked first, out of five or more models
1
In the top three
Cursor May 2026
Composer 2.5
Closed weights
No Epoch Capabilities Index score yet
- Agents' Last Exam 20.4% #19 /31
- CursorBench 4.0 27.7% #21 /21
- CursorBench 3.1 63.2% #3 /5
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SWE-bench Multilingual | 79.8% | #16 of 73 | Lab model card | |
| FrogsGame post-training task | 1.40% | #9 of 17 | Official leaderboard | |
| Vibe Code Bench | 49.6% | #65 of 120 | Official leaderboard | |
| CursorBench 3.1 | 63.2% | #3 of 5 | Lab model card | |
| FrontierSWE v1 | 33.8% | #12 of 17 | Official leaderboard | |
| ReactBench | 17.4% | #27 of 28 | Official leaderboard | |
| CursorBench | 27.7% | #21 of 21 | cursor.com, via Epoch AI | |
| CursorBench 4.0 | 27.7% | #21 of 21 | Official leaderboard |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Terminal-Bench 2.0 | 69.3% | #7 of 63 | Lab model card |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Agents' Last Exam | 20.4% | #19 of 31 | Official leaderboard |
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.