FrontierThe story, in brief
Poolside Releases Laguna S 2.1, an Open-Weight Agentic Coding Model Punching Above Its Weight Class on SWE-Bench Multilingual
118B parameters, 8B active. Poolside's Laguna S 2.1 just beat models 10x its size on coding benchmarks.
The KeyNews take
Why it matters
Open-weight coding models are rapidly closing the capability gap with massive closed models, threatening the economic moat of proprietary AI labs on developer workflows.
The key facts
6 to knowLaguna S 2.1: 118B total parameters, 8B active per token (MoE architecture)
1M token context window
Matches or beats larger models on SWE-Bench Multilingual (agentic coding)
Runs on single NVIDIA DGX Spark
Open-weight under OpenMDW-1.1 license
Developed by Poolside
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: Poolside has released Laguna S 2.1, a 118B open-weight Mixture-of-Experts coding model with 8B active parameters per token and a 1M-token context. It matches or beats models several times its size on agentic coding benchmarks, ships under OpenMDW-1.1, and runs on a single NVIDIA DGX Spark. The post…