Nanbeige4.1-3B: A Small General Model that Reasons, Aligns, and Acts

Benchmark Model Rank Results
agents-on-gaia-text-only-103-validation-tasksNanbeige4.1-3B (w/ tools, text-only)#3Accuracy: 69.90
instruction-following-on-arena-hard-auto-v2Nanbeige4.1-3B#3Accuracy: 73.2
agents-on-bfclv4Nanbeige4.1-3B#10Accuracy: 56.50
instruction-following-on-multichallengeNanbeige4.1-3B#13Accuracy: 52.21
code-generation-on-livecodebench-v6Nanbeige4.1-3B#23Pass@1: 76.9
agents-on-benchNanbeige4.1-3B#26Accuracy: 48.57
deep-research-agents-on-browsecompNanbeige4.1-3B (w/ tools)#62Accuracy: 19.12