Cognition launches SWE-2: frontier-level coding performance at up to 70% lower cost
silasalberti · x · 2026-09-10
Cognition officially introduced SWE-2, its coding model closest yet to the frontier.
- Scores on par with recent frontier models on leading evals, at up to 70% lower cost
- The team scaled RL to multiple trillions of parameters with a refined recipe that pushes the Pareto curve on both capabilities and cost
More from coding & agent
- Agent-built small classifier labels 191K documents for $0.70, vs $13–26 with batch LLMs — vanstriendaniel · 2026-09-11
- A minimal agent memory harness: five tools, Markdown, one LLM — no frameworks — dfinke · 2026-09-11
- Microsoft MVP Demos Personal AGI: Five Tools, Markdown and One LLM — dfinke · 2026-09-11
- Live Experiment: Building a Personal AGI with a Markdown Knowledge Library — dfinke · 2026-09-11
- Maestro diagnoses AI coding workflow issues with 25 targeted commands — tom_doerr · 2026-09-11
- CMU Researchers Unveil TAHI, a Human-Agent Interaction Framework for Expert-Grade AI Output — EchoShao8899 · 2026-09-11