OPEN LINK ↗
// 1h agoMODEL RELEASE
Shanghai AI Lab releases Intern-S2-Preview-397B
Shanghai AI Lab has released Intern-S2-Preview-397B, an Apache-2.0 licensed, open-weight scientific multimodal Mixture-of-Experts model built on Qwen3.5-MoE. The model features 397 billion parameters (activating approximately 17 billion per token) and is designed for advanced scientific reasoning and long-horizon agent tasks.
// ANALYSIS
While trillion-parameter models have dominated headlines, Intern-S2-Preview-397B demonstrates that high-expert MoE architectures can deliver state-of-the-art scientific and multimodal performance with highly efficient active parameter execution.
- –Highly Efficient Execution: By activating only ~17B parameters out of 397B, it dramatically reduces computational overhead during inference.
- –Scientific Focus: The model is specifically optimized for complex reasoning, tool use, and long-horizon planning in scientific domains.
- –Open-Source Accessibility: Released under the permissive Apache-2.0 license, it provides the research community with a powerful foundation model for advanced academic and scientific tasks.
// TAGS
aimachine-learningllmmoeopen-sourcemultimodalscientific-reasoningshanghai-ai-lab
DISCOVERED
1h ago
2026-07-17
PUBLISHED
1h ago
2026-07-17
RELEVANCE
8/ 10
AUTHOR
TeksEdge