Laguna S 2.1: A Compact AI Model That Outperforms Giants in Coding

Laguna S 2.1

We are thrilled to introduce Laguna S 2.1, a powerful 118B parameter model that punches well above its weight class. Despite its compact size, it rivals massive models on long-horizon coding benchmarks while supporting a 1M token context window. We developed this model in under nine weeks and are releasing full evaluation trajectories to ensure complete transparency in our rigorous testing process.

"Laguna S 2.1 is, as far as we can measure, the most capable agentic coding model in its weight class by a wide margin."