Poolside AI has released Laguna S 2.1, an 118B parameter Mixture-of-Experts model with 8B activated parameters and a 1M token context window. The model was developed in under nine weeks and demonstrates strong performance on long-horizon coding benchmarks, outperforming models many times its size in its weight class. Laguna S 2.1 achieved a 70.2% score on Terminal-Bench 2.1 and a 40.4% score on DeepSWE v1.1, with all evaluation trajectories made publicly available. AI
IMPACT This compact, high-context model could enable more sophisticated AI agents to run locally, potentially accelerating development of specialized AI tools.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Hacker News — AI stories ≥50 points →
- Claude fable
- Inkling
- Kimi k3
- Muse Spark
- Nemotron 3 Ultra
- Poolside AI
- Qwen
- SWE Atlas
- SWE-Bench Multilingual
- SWE-Bench Pro
- Terminal-Bench 2.1
- Toolathlon Verified
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
