Search papers, labs, and topics across Lattice.
This paper introduces Laguna M.1 (225.8B parameters) and Laguna XS.2 (33.4B parameters), two new Mixture-of-Experts models designed for long-horizon, agentic coding tasks. The authors detail their "Model Factory," an integrated system for efficient model development, and describe the end-to-end training process, from pre-training to quantization. Benchmarking on SWE-bench and Terminal-Bench shows that Laguna M.1 and XS.2 achieve competitive performance compared to other open models of similar size.
Agentic coding gets a boost: Laguna XS.2, a 33B parameter Mixture-of-Experts model, is now available under Apache 2.0, offering a powerful and accessible tool for long-horizon tasks.
We present Laguna M.1 and Laguna XS.2, two Mixture-of-Experts foundation models built for long-horizon, agentic coding: M.1 has $225.8$B total parameters ($23.4$B activated per token) and XS.2 has $33.4$B total ($3$B activated). Both models were trained from scratch end-to-end inside the same internal system that we refer to as our Model Factory: a tightly-integrated stack of versioned data, training, evaluation, and inference components that turn model development into an industrial process. We describe the principles and design choices of the Model Factory and also detail the end-to-end training process of our models, throughout pre-training data and architecture, post-training stages, evaluation, and quantization. On agentic software engineering and terminal benchmarks (SWE-bench Verified, SWE-bench Multilingual, SWE-Bench Pro, and Terminal-Bench 2.0) M.1 and XS.2 are competitive with state-of-the-art open models in their respective weight classes. Laguna XS.2 weights are released under Apache~2.0 at https://huggingface.co/collections/poolside/laguna-xs2.
