Poolside releases Laguna S 2.1 with 1M context and agentic focus · 2 sources tracked

Poolside has released Laguna S 2.1, a Mixture-of-Experts model with 118 billion total parameters and 8 billion activated parameters per token. This model is designed for agentic coding and long-horizon tasks, featuring a 1 million token context window and interleaved attention mechanisms. Laguna S 2.1 supports native reasoning and speculative decoding, with quantized variants available for efficient serving. AI

IMPACT This model's large context window and agentic capabilities could enable more complex, long-horizon AI applications.

RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=2 ai=1.0]

Read on Mastodon — fosstodon.org →

• Poolside
• Docker
• Google Colab
• Kaggle
• OpenAI
• OpenRouter
• poolside/Laguna-S-2.1
• SGLang
• transformers
• Vercel AI Gateway
• vLLM

• model release
• product

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →