Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Mathematical Reasoning on NuminaMath (val)
Loading...
21.7
Accuracy
Claude 3.7 Sonnet
2.98
7.84
12.7
17.56
Feb 6, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
Claude 3.7 Sonnet
Model Size=~100B+, Tra...
2026.02
21.7
Claude 3.5 Haiku
Model Size=~20B+, Trai...
2026.02
16.3
Gold Match (Verifiable Rewards)
Model Size=4B, Trainin...
2026.02
16.1
DeepSeek-R1
Model Size=37B, Traini...
2026.02
15.5
Rubric-Augmented Classifier
Model Size=4B, Trainin...
2026.02
15.2
Baseline Classifier
Model Size=4B, Trainin...
2026.02
10.5
Qwen3-4B (No RL)
Model Size=4B, Trainin...
2026.02
5.5
Mistral-7B
Model Size=7B, Trainin...
2026.02
3.7
Feedback
Search any
task
Search any
task