# SWE-2 Benchmark Scores & Performance

> SWE-2 by Cognition has 5 source-displayable benchmark rows but no public overall score. It remains unranked.

Canonical page: https://benchlm.ai/models/swe-2

Last updated: September 10, 2026

## Model Details

| Property | Value |
|----------|-------|
| Creator | Cognition |
| Source Type | Proprietary |
| Reasoning Type | Reasoning |
| Context Window | 1M |
| Overall Score | Coming soon |
| Overall Rank | Unranked |

## Family & Coverage

- Family: SWE
- Variant: base
- Benchmarks covered: 5 of 434
- Sibling models: [SWE-1.7](/models/swe-1-7)
- Related earlier model: [SWE-1.7](/models/swe-1-7)
- Coverage note: BenchLM currently has partial benchmark coverage for this model, so the overall score is conservative.

## Agentic Benchmarks

| Benchmark | Score |
|-----------|-------|
| [Terminal-Bench 2.1](/benchmarks/terminalbench21) | 92.8% |
| [Terminal-Bench 4.0](/benchmarks/terminal-bench-4) | 27.30% |

## Coding Benchmarks

| Benchmark | Score |
|-----------|-------|
| [FrontierCode 1.1 Main](/benchmarks/frontiercode) | 50.0% |
| [Terminal-Bench 2.1](/benchmarks/terminalbench21) | 92.8% |
| [DeepSWE](/benchmarks/deepswe) | 73.0% |

## Other Cognition Models

- [SWE-1.7](/models/swe-1-7) - Score: not computed