In the news
Introducing SWE-2 Flash
Cognition · Published · 3 min read
In 30 seconds
- What happened
- Cognition released SWE-2 Flash, a smaller coding model achieving 20% on Terminal-Bench 4.0, up from 0.9% baseline, at one-fifth the cost.
- Why it matters
- Engineers building software agents, automations, and code review tools who need faster inference and lower costs than larger models.
- Watch out
- Weights are not publicly available; access requires approval from Cognition's partner program due to cybersecurity concerns and potential misuse risks.
The patterns behind this
Each one covers how the technique works, when it earns its cost, and where it breaks.
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.