by Poolside AI
Poolside's open-weight Laguna 2.1 coding models: S (118B MoE, 8B active, 1M context) and XS (33B MoE, 3B active).
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Laguna S 2.1 is a mixture-of-experts model from Poolside AI with 118B total parameters, 8B active per token and a 1M-token context window. Poolside positions it for reasoning and agentic coding at mid-size cost. It belongs to the Laguna family, which began with Laguna M.1 and XS.2, released in April 2026 alongside the runtime Poolside uses to train and operate agents.
Laguna XS 2.1 is the smaller variant, with 33B total parameters, 3B active and a 256K context window. Poolside describes it as its lightest and fastest agentic coding model and says it is small enough to run on-device. Poolside's site describes the models as open-weight, but the scraped pages did not give a license or download location.
A YouTube creator reported that S 2.1 scored slightly above GLM 5.2 on Deep Sweep, and that XS 2.1 performed worse than Qwen 3.6 in their own test. These are single third-party observations, not verified benchmarks. Pricing, API access and licensing details were not found in the provided material.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Open-weight 118B MoE coding model with 8B active parameters and 1M context, plus a smaller 33B XS variant. Scores slightly above GLM 5.2 on Deep Sweep; the XS version was weaker than Qwen 3.6 in the creator's test.
Closest neighbours by meaning, not just by category.
Model · IQuest Lab
320B-parameter MoE model (15B active) from IQuest for agentic coding and tool use, with 512K context.
Model · Recursive Intelligence
Dense 65B open-weight language model from Ricursive Intelligence, trained for agentic coding with up to 1M context.
Model
35B Mixture-of-Experts open-weights model from InternScience for long-horizon agentic search, engineering and tool use.
Model
Open-weights family of agentic coding models (9B to 397B) that learns to generate its own task-specific scaffolds.
Model · BAAI
27B open-weights agent model from BAAI that iteratively proposes, tests and refines solutions on coding and ML tasks.
Model
Open model family (9B dense, 35B MoE, 397B MoE) trained with a self-improvement loop for reasoning, coding and agentic tasks.