Intern-S1

Intern-S1 is a scientific multimodal foundation model released and open-sourced by the Shanghai AI Laboratory at WAIC 2025 on July 26, 2025. Built on Qwen3-235B MoE (with an Intern-S1-mini on Qwen3-8B), it adds a cross-modal scientific analysis engine that interprets chemical formulas, protein structures, seismic signals, and other scientific data, plus an Intern-Discovery platform for researcher collaboration.

It claims state-of-the-art performance among open-source models on scientific reasoning benchmarks (ChemBench, MatBench, ProteinLMBench), competitive with leading closed systems such as o3 and Gemini-2.5-Pro, while holding top-tier general reasoning scores. Model weights and toolchains are open-sourced, making it the first open general model to integrate professional scientific capabilities. Vendor benchmark claims should be treated as vendor-reported until independently reproduced.