Cognition's SWE-2 hits 92.8 on Terminal-Bench 2.1

cdnsteve · hn · 2026-09-11

A Hacker News post links to a report that Cognition's SWE-2 scored 92.8 on Terminal-Bench 2.1, placing it at the top of the agentic coding benchmark. The post itself offers no further details beyond the link to the model page.

Original post →

More from Models

Models channel →