# Laguna S 2.1 is live on Kilo > Source: > Published: 2026-07-21 19:24:30+00:00 # Laguna S 2.1 is live on Kilo ### Poolside's latest model is free for a limited time Poolside just [shipped Laguna S 2.1](https://poolside.ai/blog/introducing-laguna-s-2-1), and you can run it in Kilo today across the IDE extension, CLI, and Cloud Agents. Try it for free for a limited time! [Poolside](https://poolside.ai/) builds open-weight foundation models and they share Kilo’s commitment to building in the open. Their previous models have been super popular with developers at the cutting edge, and we’re excited for our community to try [the new Laguna S model](https://kilo.ai/models/poolside-laguna-s-2-1). Here’s why Laguna S 2.1 is worth your attention: it’s a 118B parameter model that only activates 8B per token, and it’s already going head to head with models many times its size. On SWE-bench Multilingual it hits 78.5%, which puts it right in the mix with [Tencent Hy3](https://kilo.ai/models/tencent-hy3) (295B-A21B) and DeepSeek-V4-Pro Max (1.6T-A49B). On Toolathlon Verified it scores 49.7%, ahead of Nemotron 3 Ultra and DeepSeek-V4-Flash Max. That’s a lot of capability packed into a model small enough to run locally on a single NVIDIA DGX Spark. ## The thing that actually matters: it doesn’t quit Benchmarks are one thing, but the more interesting story with Laguna S 2.1 is behavioral. Poolside reports it holding onto a task, using tools, checking its own work, and recovering from failed approaches for up to 24 hours with little to no intervention. That’s the kind of long-horizon persistence that separates a model you babysit from one you actually delegate to. A few examples from Poolside’s internal testing give a sense of what that looks like in practice: It built a full HTML and CSS rendering engine over 181 reasoning turns, comparing its own output against a real browser pixel by pixel until it landed within three pixels of the reference. In a restricted sandbox with no internet access, it independently rediscovered a known solution to Erdős Problem 397, and when Python wasn’t available, it switched to Perl and derived a distinct parametric family on its own. It also spent an overnight run improving on a circle-packing result from DeepMind’s AlphaEvolve, and kept nudging the number closer to optimal after the first improvement instead of stopping. None of that reads like a demo script. It reads like a model that was just left to work. ## Why this fits Kilo Kilo runs on open pricing and open model selection, which means the second a model like Laguna S 2.1 lands, you can point your existing sessions at it and pay exactly what Poolside charges per token, nothing more. No waiting on a plan tier, no separate integration. That matters more for a model like this than most. If [Laguna S 2.1](https://kilo.ai/models/poolside-laguna-s-2-1) is genuinely built for long-horizon agentic work, the place to feel that is KiloClaw, where it can run proactively in the background without you babysitting every step, or in Cloud Agents and Parallel Agents, where you can let it churn on a real task while you work on something else. The efficient 8B-active parameter count also means those long runs and RL-style iteration loops stay cheap, which is the whole point of open model selection: pick the model that fits the job, not just the one everyone defaults to. ## Try it today! [Laguna S 2.1](https://kilo.ai/models/poolside-laguna-s-2-1) is available now in the [IDE extension, CLI, and Cloud Agents](https://kilo.ai/). Switch to it from your model picker, or point KiloClaw at it for background work. Check out the full model card and benchmarks [on Hugging Face](https://huggingface.co/poolside/Laguna-S-2.1) and watch the [Kilo Leaderboard](https://kilo.ai/leaderboard) to see if it claims a top position like Laguna M.1 did.