Poolside AI Launches 118B Laguna S2.1 with a 1M Context Window

Poolside AI released Laguna S2.1, an open-source LLM.

The model uses a 118B-parameter mixture-of-experts design and a 1-million-token context window.

It includes dual reasoning modes and NV4 quantization, enabling local use on consumer GPUs.

Its long context and efficient deployment could reduce the cost of advanced AI workflows.