Cognition launches SWE-2 Flash, a 276B-A12B coding model that beats all US open-weight rivals
carlesgelada · x · 2026-11-05
Cognition has released SWE-2 Flash, a new fast and efficient coding model (276B-A12B MoE).
- On official benchmarks it outperforms all American open-weight models, even ones 3x larger: FrontierCode 2.0 24.0, Terminal-Bench 2.1 89.2, SWE-Bench Pro v2 Hard 89.9.
- It trades some performance vs the 2.8T SWE-2 (e.g. 20.0 vs 27.3 on Terminal-Bench 4.0) for speed, while still rivaling GPT-6 Luna and Claude Haiku 5.5 on several evals.
- To balance transparency and cyber risk, Cognition is building a program to securely share the weights with partners and researchers.
Related event: Cognition Releases SWE-2 Flash, Outperforming Larger US Open-Weight Models(2 posts)→
More from coding & agent
- LMArena opens Agent Arena for agent evaluation — arena · 2026-11-05
- Running Qwen 27B Q4 at 0.7 tok/s on a 4GB-VRAM laptop as a main coding model — Felix_455-788 · 2026-11-05
- His agent now auto-labels his Google Calendar in the background — steren · 2026-11-05
- The security startups racing to rein in AI agents — pstAsiatech · 2026-11-05
- OpenAI October dev roundup: Decisions API, 12-hour caching, 8x faster Codex tokens — OpenAIDevs · 2026-11-05
- Datadog previews Canvas, a new way to collaborate with AI agents on observability — farooqib · 2026-11-05