Cognition SWE-2 Flash Release - ExploitBench: leaderboard
ExploitBench exploit-development evaluation; the post does not state the target set or budget. Cognition's SWE-2 Flash launch post (2026-11-04), as reported by Cognition for its own SWE-2 Flash and SWE-2 only; the post states no harness, attempts or reasoning effort, and its comparator columns are not admitted. Score (%), higher is better.
Metric: Score (%). Source: cognition.com. Saturation forecast: Rough model projection: around 2026. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | SWE-2 | 36.1 | |
| 2 | SWE-2 Flash | 34.8 |
Interactive version: theaggregate.ai/benchmark?slug=cognition-swe-2-flash-release-exploitbench · How It Works · Data refreshed daily, snapshot 2026-11-07.