Laguna XS 2.1
Laguna XS 2.1 updates XS.2. It has 33 billion total and 3 billion active parameters, with a mixture of global and sliding-window attention.
What the release provides
The text-only release preserves reasoning between tool calls and allows thinking to be switched off. FP8 context-cache storage reduces memory per token. Official lower-precision packages support local deployments. The license is OpenMDW 1.1.Poolside: Laguna XS 2.1 model card
How to use it well
Practical guidance. Check sustained memory use at the context length you need. A model that loads on a device can still exceed memory limits during a long tool conversation.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable, Provider product |
| Model identifiers | poolside/Laguna-XS-2.1 |
| Context window | 262,144 tokens Published model context window Poolside: Laguna XS 2.1 model card |
| Supported deployment | Included in Poolside’s self-managed inference offeringPoolside: Supported models |
Related models
- Laguna S 2.1
Poolside · Laguna
An open-weight coding model with a one-million-token context.
Launch date not confirmedDownloadable · Provider product - Laguna XS.2
Poolside · Laguna
The earlier open-weight small Laguna coding model.
Launch date not confirmedDownloadable