Model library / poolside
Laguna S 2.1 is a 118 billion-parameter mixture-of-experts model for agentic coding with long-context support.
Parameters in total
Context length in tokens, as published
Weights at the released precision
Licence declared on the repository
Overview
Laguna S 2.1 is a 118 billion total-parameter mixture-of-experts language model released by poolside. It uses a token-choice router with softplus gating over 256 routed experts plus one shared expert, grouped-query attention, and a mixed global and sliding-window attention layout. The model is targeted at software engineering and agentic coding, offering native interleaved reasoning and tool-calling capabilities.
In practice the model comprises 48 layers, 12 global and 36 sliding-window attention layers, and activates roughly 8 billion parameters per token. It supports a 1,048,576-token context window. The BF16 checkpoint occupies about 236 GB, requiring multiple GPUs, while quantised variants reduce memory needs. It is released under the permissive OpenMDW-1.1 licence, allowing unrestricted commercial use.
Specification
| Repository | poolside/Laguna-S-2.1 |
| Publisher | poolside |
| Published | 13 July 2026 |
| Architecture | LagunaForCausalLM |
| Model type | laguna |
| Parameters | 117,561,977,600 |
| Layers | 48 |
| Hidden size | 3,072 |
| Attention heads | 48, 8 key/value heads |
| Context length | 1,048,576 tokens |
| Vocabulary | 100,352 tokens |
| Weight format | bfloat16 |
| Licence | openmdw-1.1 |
Memory
Parameter count multiplied by the bytes each format uses. The key-value cache comes on top.
| Precision | Weights | Notes |
|---|---|---|
| bf16 | 219.0 GiB | Full precision as released |
| fp8 | 109.5 GiB | 8-bit, near-lossless on most models |
| int4 | 54.7 GiB | 4-bit, smallest footprint |
The cache costs 192 KiB per token at 16-bit, so the full 1,048,576-token context of one request needs about 192.0 GiB. Concurrency multiplies that number, not the weights.
That figure is an upper bound: this model uses sliding-window attention with a 512-token window on some layers, so long requests cache less than this.
Hardware
At the released precision, with 90% of the card given to the server and 2.5 GiB kept for the runtime. The context column is what is left for the cache on that setup.
| GPU | Cards needed | Free for cache | Context it holds |
|---|---|---|---|
| RTX 4090 24GB | more than 8 | — | — |
| RTX 5090 32GB | more than 8 | — | — |
| L40S 48GB | 8 | 106.6 GiB | about 582,307 tokens |
| A100 80GB | 4 | 59.0 GiB | about 322,348 tokens |
| H100 80GB | 4 | 59.0 GiB | about 322,348 tokens |
| RTX PRO 6000 96GB | 4 | 116.6 GiB | about 636,921 tokens |
| H200 141GB | 2 | 29.8 GiB | about 162,877 tokens |
Serving
A vLLM launch line for the shape above. Check the model card for a runtime the publisher recommends.
vllm serve poolside/Laguna-S-2.1 \ --tensor-parallel-size 8 \ --max-model-len 581632 \ --gpu-memory-utilization 0.90
Questions
The weights take about 219 GiB at the released precision. No single card in the table above holds that, so it needs several GPUs or a lower precision.
The repository declares openmdw-1.1. Read the licence text before commercial use: the name alone does not tell you what is allowed.
The configuration allows 1,048,576 tokens. Whether the whole window is usable depends on the memory left for the cache, which the hardware table works out per card.
This page is a reference, and Aiqre does not serve this model today. We deploy open-weight models on dedicated EU hardware on request, so if you want an OpenAI-compatible endpoint for it, email [email protected] with the model name and roughly what volume you expect.
Also in the library
Ling-3.0-flash-Fin is a 127.5 billion-parameter finance-enhanced language model for long-context financial research and analysis.
Ling-3.0-flash is a 127.5 B parameter hybrid-linear language model from inclusionAI for reasoning and long-context applications.
Hosting
Tell us the model and roughly what volume you expect. We reply with a price and an OpenAI-compatible endpoint, with your prompts never stored.