| instruction-following-on-arena-hard-auto-v2 | Nanbeige4-3B-Thinking-2511 | #4 | Accuracy: 60.0 |
| mathematical-reasoning-on-aime24 | Nanbeige4-3B-Thinking-2511 | #10 | Acc: 90.4 |
| agents-on-bfclv4 | Nanbeige4-3B-Thinking-2511 | #11 | Accuracy: 53.8 |
| code-generation-on-livecodebench-v5 | Nanbeige4-3B-Base + OpenThoughts3 SFT (extra data) | #15 | Pass@1: 40.9 |
| instruction-following-on-multichallenge | Nanbeige4-3B-Thinking-2511 | #15 | Accuracy: 41.2 |
| multiple-choice-question-answering-mcqa-on-supergpqa | Nanbeige4-3B-Thinking-2511 | #15 | Accuracy: 53.2 |
| mathematical-reasoning-on-math-500 | Nanbeige4-3B-Base + Ring-Lite SFT (extra data) | #22 | Accuracy: 95.5 |
| mathematical-reasoning-on-aime-2025 | Nanbeige4-3B-Thinking-2511 | #32 | Accuracy: 85.6 |
| multiple-choice-question-answering-mcqa-on-gpqa-diamond | Nanbeige4-3B-Thinking-2511 | #50 | Accuracy: 82.2 |
| code-generation-on-livecodebench-v6 | Nanbeige4-3B-Base + OpenThoughts3 SFT (extra data) | #58 | Pass@1: 38.8 |