Poolside launches Laguna S 2.1, an 118B open-weight model built for agentic coding and long-horizon work, that it says competes with larger open models
Chinese open-source models like Kimi, Deepseek and Qwen are now dominant AI tools. American upstarts like Mira Murati's Thinking Machines …
ForbesIain Martin
Context & Ripple Effects
Poolside’s new release extends its move from the smaller open-weight Laguna XS.2 alongside a proprietary model into a substantially larger open-weight offering. The company is targeting coding agents and tasks that require sustained work rather than a general-purpose positioning alone.
The launch arrives amid a visible contest around open-weight, long-horizon coding models: Moonshot had already positioned Kimi K2.6 around improved long-horizon coding capabilities. Poolside’s claim of competing with larger models makes model size efficiency a central part of its pitch.
First-order effects
Developers and organizations evaluating coding agents gain another 118B open-weight model option, while Poolside can test its competitiveness directly against larger open models on the workloads it is targeting.
Poolside shifts its public product line toward larger open weights after its earlier 33B-A3B Laguna XS.2 release, giving its open-model strategy a more capable flagship.
Second-order effects
Open-model vendors focused on coding and agentic work face added pressure to demonstrate performance, efficiency, and usable long-horizon behavior rather than rely on parameter scale alone.
Buyers comparing open models for coding can gain negotiating leverage as more vendors offer alternatives aimed at the same specialized workloads, though real leverage depends on independent deployment and performance results.
Third-order effects
If comparable capability continues moving into open-weight models, value may shift from access to a base model toward deployment, tooling, evaluation, and workflow integration around it.
The market is increasingly segmenting around task-specific agent performance—especially coding and sustained execution—rather than a single leaderboard for general model capability.
The trend: Open-weight AI competition is moving toward efficient, workload-specific models designed to power agentic software development.
Today we are releasing Laguna S 2.1. At 118B total parameters, with 8B active per token, it does the work of models several times its size on agentic coding. It is remarkably persistent across long-horizon tasks. And it is small enough to run on a single NVIDIA DGX Spark. It is
Quite excited about this one: Laguna S 2.1: 118B-A8B MoE with 1M context. It runs on a single NVIDIA DGX Spark, and it stacks up very well on what we focus on: agentic coding. A few observations we had building it. 🧵
I'm very curious about this one. Like I been running 2 x 3090s on Qwen 3.6 27b and 2 x 3090 on Qwen 3.6 35B A3.... A some point I got to move on right... Would this be solid to run on all 4 ? It can fit.
Big day for American open-source AI. For the launch of Laguna S, I sat down with @eisokant to discuss its architecture, the economics of open weights, and the question of who gets to build intelligence. Timestamps: 0:00 Intro 1:50 Why Poolside started opening its models: the [vid…
52 days from kickoff to a 118B/8B MOE beating models 4-25x its size isn't a benchmark flex, it's a claim about the factory that made it. Publishing the trajectories instead of just the scorecard is the right instinct - lets go & find the holes.
the more i look into it, the more impressive it is! best in class:) notable things were fp8 training for RL and scaling RL with big focus on terminal use related tasks!! [image]
I expect US open source models to come out in droves over the next 12 months. Even most of the close source labs will likely release open source models to compete for market shares of the explosively growing lower-end enterprise inference market.
Poolside was an early AI star raising a $500 million round back in 2024 but then largely disappearing from view. Now it has launched a big new open-source AI model it claims can beat China's best https://www.forbes.com/...
Today we're releasing Poolside Laguna S 2.1 It is a 118B-total, 8B-active open-weight model built for agentic coding and long-horizon work, with context up to 1M tokens https://poolside.ai/... Laguna S 2.1 sits at the top of its weight class and competes with open models many
Poolside Laguna S2.1; a step up from before. 40+ on DeepSWE is a good number. But perhaps svg and html aren't where it shines. We shouldn't judge all model on same standard. [image]