poolside releases Laguna M.1 for agentic coding work
Laguna M.1 is a 225 billion parameter model from poolside. It does not use the whole model for every token; it activates about 23 billion parameters per token, which points to a design meant to keep large-model capability while reducing compute load. The model is built for agentic coding and long-running tasks.
It has 70 layers: 3 dense SwiGLU layers first, followed by 67 sparse MoE layers that route work across 256 experts. It uses global attention in every layer and supports reasoning between tool calls, with reasoning turned on or off for each request. poolside presents it as competitive with leading open-weight and on , SWE-bench Multilingual, SWE-Bench Pro, and .
It is released under an , so it can be used and modified for both commercial and non-commercial work. Its training included pre-training, post-training, and stages.
Key points
- The model has 225 billion total parameters but uses about 23 billion per token.
- It is aimed at agentic coding and .
- Its sparse MoE design routes work through selected experts instead of using the whole model every time.
- It supports reasoning between tool calls and lets that reasoning be enabled or disabled per request.
- The allows commercial and non-commercial use.