List your company

Mathematical and General Reasoning on DeepMATH (test)

83.4MATH 500 ScoreBF16
56.77663.68870.677.512Jan 20, 2026

Evaluation results

MethodLinks
BF16Model=Qwen3-8B-Base, Rollout Length=16k, Training Dataset=DeepMATH2026.0183.4—44.236.154.6
Jet-RLModel=Qwen3-8B-Base, Rollout Length=16k, Training Dataset=DeepMATH2026.0180.2—47.233.853.7
Before TuningModel=Qwen3-8B-Base, Rollout Length=16k, Training Dataset=DeepMATH2026.0169.763.446.231.852.8
BF16-Train-FP8-RolloutModel=Qwen3-8B-Base, Rollout Length=16k, Training Dataset=DeepMATH2026.0157.8—42.632.644.3