List your company

Mathematical Word Problems on GSM8K (Nec, Suf, Shuf)

88.7NecNemotron-Ultra
-2.09221.47945.0568.621Mar 24, 2026

Evaluation results

MethodLinks
Nemotron-UltraN=306–498, Pattern=Ctx Dep†2026.0388.711.190.8
Small models (0.8–8B avg)N=100 ea., Pattern=Task-split2026.035514—
MiniMax-M2.5N=427–496, Pattern=Genuine2026.0328.470.526.8
GPT-OSS-120BN=425–496, Pattern=Decorative2026.0310.988.99.4
GPT-5.4N=376–500, Pattern=Decorative2026.038.880.32.5
Claude Opus 4.6-RN=486–499, Pattern=Decorative2026.034.888.72.4
DeepSeek-V3.2N=500, Pattern=Decorative2026.033.693.42.7
Kimi-K2.5N=457–500, Pattern=Decorative2026.031.490.91.5