Laguna S 2.1 Free

Laguna S 2.1 Free is Poolside's purpose-built coding agent model, aimed squarely at software engineering and agentic workflows where models need to plan, use tools, and iterate on real codebases. It pairs text in and out with reasoning and tool calling, the two capabilities that matter most when an assistant has to read long files, decide on a sequence of edits, and verify results. That developer-first posture shows up across third-party listings, which consistently frame it as one of the stronger coding models in its weight class.

Under the hood, the model uses a Mixture-of-Experts design with 118B total parameters and 8B active per pass, letting it stay efficient while keeping a large knowledge store for programming tasks. OpenRouter cites 70.2% on Terminal-Bench 2.1 and 40.4% on DeepSWE as evidence of competitive agentic coding ability, and the weights are openly published on Hugging Face under the OpenMDW-1.1 license. A roughly 262K-token context window with around 33K tokens of output supports deep analysis across large repositories, while zero-cost input and output on the free tier make it approachable for experimentation and long-doc workflows.

Vercel AI Gatewaypoolside/laguna-s-2.1-freelaguna

Open provider page Provider docs

Quick Info

Provider: Vercel AI Gateway
Model key: poolside/laguna-s-2.1-free
Release date: Jul 21, 2026
Last updated: Jul 21, 2026
Input modalities
Output modalities
Capabilities

Cost

A provider subscription or plan supersedes token-based pricing for this model.

Limits

Output tokens: 32,768 tokens
Context window: 256,000 tokens

Latest news about Laguna S 2.1 Free

Vercel AI Gateway

Official sourceRelease Notes

Vercel announced on July 31, 2026 that poolside/laguna-s-2.1 and poolside/laguna-s-2.1-free each received a 10x capacity increase on AI Gateway, allowing both paid and free endpoints to handle far more concurrent requests. The changelog positions the expansion as enabling high-volume agentic coding workloads and long-r Developers can adopt the higher-capacity endpoints immediately via the AI SDK by setting the model identifier to poolside/laguna-s-2.1-free or poolside/laguna-s-2.1, or by running vercel ai-gateway coding-agents setup and selecting either model in the agent's configuration. The upgrade applies to both SKUs and inherits

Vercel AI Gateway

Coverage

Poolside released Laguna S 2.1 on July 22, 2026, as an open-weight coding model built on a Mixture-of-Experts architecture with 118 billion total parameters but only 8 billion active per token. According to the report, the model supports a context window of up to one million tokens and offers separate thinking and no-t The release highlights a 70.2% score on Terminal-Bench 2.1 and a memory footprint of roughly 71GB, with reported inference speeds of 50–77 tokens per second on a single DGX Spark. Poolside states the training run completed in under nine weeks on 30 trillion tokens under a permissive open-weight license. The same day, O