{"ID":2873791,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.06079","arxiv_id":"2509.06079","title":"Multimodal Reasoning for Science: Technical Report and 1st Place Solution to the ICML 2025 SeePhys Challenge","abstract":"Multimodal reasoning remains a fundamental challenge in artificial intelligence. Despite substantial advances in text-based reasoning, even state-of-the-art models such as GPT-o3 struggle to maintain strong performance in multimodal scenarios. To address this gap, we introduce a caption-assisted reasoning framework that effectively bridges visual and textual modalities. Our approach achieved 1st place in the ICML 2025 AI for Math Workshop \\\u0026 Challenge 2: SeePhys, highlighting its effectiveness and robustness. Furthermore, we validate its generalization on the MathVerse benchmark for geometric reasoning, demonstrating the versatility of our method. Our code is publicly available at https://github.com/OpenDCAI/SciReasoner.","short_abstract":"Multimodal reasoning remains a fundamental challenge in artificial intelligence. Despite substantial advances in text-based reasoning, even state-of-the-art models such as GPT-o3 struggle to maintain strong performance in multimodal scenarios. To address this gap, we introduce a caption-assisted reasoning framework tha...","url_abs":"https://arxiv.org/abs/2509.06079","url_pdf":"https://arxiv.org/pdf/2509.06079v1","authors":"[\"Hao Liang\",\"Ruitao Wu\",\"Bohan Zeng\",\"Junbo Niu\",\"Wentao Zhang\",\"Bin Dong\"]","published":"2025-09-07T14:47:32Z","proceeding":"cs.CL","tasks":"[\"cs.CL\",\"cs.CV\"]","methods":"[]","has_code":false,"code_links":[{"ID":610095,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2873791,"paper_url":"https://arxiv.org/abs/2509.06079","paper_title":"Multimodal Reasoning for Science: Technical Report and 1st Place Solution to the ICML 2025 SeePhys Challenge","repo_url":"https://github.com/OpenDCAI/SciReasoner","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}