Cognition's SWE-2 hits 50% on FrontierCode, 64% cheaper than Fable 5.1
carlesgelada · x · 2026-09-12
Cognition released SWE-2, its most advanced coding model, claiming a push of the capability-cost Pareto frontier: 50.0% on FrontierCode 1.1 Main, within one point of Fable 5.1 at 64% lower cost, beating Grok 4.6 on score and cost and approaching GPT-6 Astra at a quarter of the price. Key training advance: scaling RL to the multi-trillion-parameter regime with a single-run algorithm that trains all reasoning-effort levels jointly via per-level linear cost penalties—no extra hyperparameters. SWE-2 is post-trained from Kimi K3 (2.8T params), adding 5-6 points on many benchmarks.
More from coding & agent
- Firecrawl offers 2.5x credits to a16z Speedrun founders, unveils $75M Series B — devdigest · 2026-10-27
- Chums: open-source Apple Watch app puts Claude Code and Codex on your wrist — brandon_galang · 2026-10-27
- Open-source Clairic Studio turns pasted text into typed data and working code in 5 languages — dfinke · 2026-10-27
- Matt Pocock shares expand/contract trick: split one-way-door PRs to de-risk big changes — mattpocockuk · 2026-10-27
- Qdrant builds Universal I/O to run vector search mmap code on object storage — qdrant_engine · 2026-10-27
- theo publishes tstack skills, Garry Tan urges everyone to 'publish your stack' — ycombinator · 2026-10-26