PoolsideLaguna
Laguna M.1
Laguna M.1 has 225 billion total and 23 billion active parameters. It uses global attention throughout its seventy layers.
What the release provides
The first three layers are dense; the rest use sparse experts. Post-training adds reasoning and tool use to the pretrained model. Apache 2.0 covers the checkpoint, with FP8 and NVFP4 packages also available.Poolside: Laguna M.1 model card
How to use it well
Practical guidance. Plan deployment around the whole checkpoint and the intended context. Compare complete task outcomes before choosing this larger model over a smaller Laguna release.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable, Provider product |
| Model identifiers | poolside/Laguna-M.1 |
| Context window | 262,144 tokens Published model context window Poolside: Laguna M.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 M.1 base
Poolside · Laguna
The pretrained M.1 checkpoint for further model training.
Launch date not confirmedDownloadable