SWE-2: benchmark results
Provider: Other. Released 2026-09-10. Access: Interface.
Aggregate Rating 1867 ± 49, rank #31 of 10870 rated models, from 13 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AI Coding Daily (Devin) - Laravel Code Quality | 17.5 | Laravel Code Quality (max 20) points, LLM-judged rubric scor | 100 |
| AI Coding Daily (Devin) - Total | 73.3 | Total points (max 90) | 100 |
| Cognition SWE-2 Flash Release - DeepSWE 1.1 | 73 | Score (%) | 100 |
| Cognition SWE-2 Flash Release - ExploitBench | 36.1 | Score (%) | 100 |
| Cognition SWE-2 Flash Release - FrontierCode 1.1 Main | 50 | Score (%) | 100 |
| Cognition SWE-2 Flash Release - SWE-Atlas QA | 56.3 | Score (%) | 100 |
| Cognition SWE-2 Flash Release - SWE-Bench Pro v2 Hard | 90.6 | Score (%) | 100 |
| Cognition SWE-2 Flash Release - Terminal-Bench 2.1 | 92.8 | Score (%) | 100 |
| Cognition SWE-2 Flash Release - Terminal-Bench 4.0 | 27.3 | Score (%) | 100 |
| LLM Stats Score | 47.53 | LLM Stats Score (conservative rating) | 91.3 |
| AI Coding Daily (Devin) - Python Code Quality | 16.73 | Python Code Quality (max 20) points, LLM-judged rubric score | 50 |
| AI Coding Daily (Devin) - React-TS Code Quality | 17.83 | React-TS Code Quality (max 20) points, LLM-judged rubric sco | 50 |
Interactive version: theaggregate.ai/model?slug=swe-2 · How It Works · Data refreshed daily, snapshot 2026-11-12.