GitLab Transcend - Intelligent orchestration
Mini toolsIntern S2 Preview
By InternLM
Intern-S2-Preview is a 35B scientific multimodal foundation model from InternLM, released on Hugging Face under Apache 2.0. It is an image-text-to-text model continued from Qwen3.5 and designed for scientific reasoning, multimodal understanding, and agent capabilities. The model targets professional scientific workflows, including specialized scientific tasks, spatial modeling for small-molecule structures, material crystal structure generation, tool calling, and long-form reasoning. The page states evaluation settings of up to 128K tokens for text reasoning and 64K tokens for multimodal benchmarks.
New
Multimodal
Gen 3
Overview
Intern-S2-Preview is InternLM’s open-source 35B scientific multimodal foundation model for image-text reasoning, scientific tasks, and agent workflows.
About InternLM
Welcome to the InternLM (Intern Large Models) organization. Intern-series large models (chinese name: 书生) are developed by Shanghai AI Laboratory and we keep open-sourcing high quality LLMs/MLLMs as well as toolchains for development and application.
Website:
internlm.intern-ai.org.cn
Tools using Intern S2 Preview
No tools found for this model yet.
Last updated: May 18, 2026
