What happened
Sage AI introduced Sage S2.1 Flash, a multimodal model for research tasks involving multiple rounds of analysis and verification. It combines scientific data understanding with information retrieval, coding, and tool use, and can check analyses against intermediate outputs and visual feedback. The model is now supported in vLLM, and its weights are available on Hugging Face and ModelScope.
Great work from @sageai_official on Sage S2.1 Flash, now supported in vLLM!
From molecular graphs to protein structures, it brings scientific data understanding together with retrieval, coding and tool use—checking intermediate results and refining the next step.
Weights are available on @HuggingFace and @ModelScope2022.
View the quoted post on X🚀 Introducing Sage S2.1 Flash, our new multimodal model designed for research tasks requiring multiple rounds of analysis and verification.
1⃣Combines scientific data understanding with information retrieval, coding, and tool use.
2⃣Uses specialized software to perform analyses, verify results against intermediate outputs and visual feedback, and adjust subsequent steps accordingly.
3⃣Built on the unified representation architecture of Sage SciPerception, supporting more than 10 types of scientific input, including molecular graphs, crystal structures, protein sequences, and 3D protein structures.
4⃣Ranks first on MLE-bench Lite (90.3), OfficeQA Pro (72.2), Harvey LAB (25.0), and SpreadsheetBench 2 (37.1).
🥰Now supported by vLLM (@vllm_project) and SGLang ( @lmsysorg ) — with more ecosystem integrations on the way.
🤗 Model: @HuggingFace huggingface.co/internlm/Sage-…
@ModelScope2022 modelscope.cn/models/Shanghai…
🤗 Try it now:
discovery.intern-ai.org.cn/to…
The story so far
Sage AI introduced Sage S2.1 Flash, a multimodal model for research tasks involving multiple rounds of analysis and verification, with support now available in vLLM. According to vLLM, the model combines scientific data understanding with information retrieval, coding, and tool use, and can check analyses against intermediate outputs and visual feedback. Its weights are available on Hugging Face and ModelScope.
