Developers after an open coding model they can run locally or call very cheaply should look here first.
Kenning's take
What it does
Laguna S 2.1 is a coding and agent model from Poolside with 118 billion total parameters, of which 8 billion are active per token, and a context window of up to 1 million tokens. Poolside released it with open weights on Hugging Face and published full benchmark trajectories alongside its scores. It can be run locally through vLLM, SGLang, Ollama or atomic.chat, or used hosted through OpenRouter, Baseten and Vercel AI Gateway, and inside coding agents such as Cline, OpenCode and Poolside's own pool.
- 118B total parameters, 8B active (Mixture-of-Experts)
- Up to 1M-token context, thinking and no-thinking modes
- Open weights under OpenMDW-1.1 in BF16, FP8, INT4 and NVFP4
- Official GGUF and MLX conversions
- Free 256K-context endpoint on OpenRouter
Who it suits. Developers who want a capable open coding model they can run on their own machines or call cheaply.
Key facts
- Category
- AI models
- Made by
- Poolside
- Released
- 21 July 2026
- How to get it
- open weights on Hugging Face; hosted through OpenRouter (free and paid endpoints), Baseten and Vercel AI Gateway; inside Cline, OpenCode, Kilo and Poolside's pool agent
- Works on
- API, self-hosted, Hugging Face
- In the UK
- Yes
- Free plan
- open weights, plus a free 256K-context endpoint on OpenRouter
- API price
- $0.10 per million input tokens, $0.20 per million output, $0.01 cache read (launch post); OpenRouter's model list showed $0.09 and $0.18 on 2026-10-11
FAQ
When was Laguna S 2.1 released?
Poolside released Laguna S 2.1 on 21 July 2026.
Is Laguna S 2.1 free?
The weights are free on Hugging Face under OpenMDW-1.1, and OpenRouter has a free endpoint with 256K context. A paid endpoint gives the full 1M context.
How big is Laguna S 2.1?
It has 118 billion total parameters with 8 billion active per token.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
ACE-Step 1.5
Open-source music generation model that writes full songs with lyrics and runs on ordinary consumer GPUs.
AlphaGenome Atlas
Google DeepMind database of AI-predicted effects for every possible single-letter change in human DNA.
Amazon Nova 2
Amazon's second family of Nova AI models, covering reasoning, speech conversation and combined text-and-image generation.
Cisco Antares
Cisco's small open-weight models that find where known security vulnerabilities sit in a codebase, running locally.