List your company
ResearchFormerly WizWand

Mathematical Reasoning on Average (Math500, AIME25, AMC23)

13,040LengthBase Model
1,898.484,790.997,683.510,576.01Mar 9, 2026
Updated 7mo ago

Evaluation results

MethodLinks
Base ModelBackbone=Qwen3-4B-Thinking-2507, #RL Steps=N/A2026.0313,04088.5—
Base ModelBackbone=DeepSeek-R1-Distill-Qwen-1.5B, #RL Steps=N/A2026.039,98060.7—
Base ModelBackbone=DeepSeek-R1-Distill-Qwen-7B, #RL Steps=N/A2026.038,46373.1—
SmartThinkerBackbone=Qwen3-4B-Thinking-2507, #RL Steps=502026.037,747890.42
Truncated Think 8kBackbone=DeepSeek-R1-Distill-Qwen-1.5B, #RL Steps=N/A2026.035,65846.7-0.72
Truncated Think 8kBackbone=Qwen3-4B-Thinking-2507, #RL Steps=N/A2026.035,50119.6-3.31
SmartThinkerBackbone=DeepSeek-R1-Distill-Qwen-7B, #RL Steps=752026.035,38274.50.43
SmartThinkerBackbone=DeepSeek-R1-Distill-Qwen-1.5B, #RL Steps=1502026.035,16961.90.54
LASER-DE-4096Backbone=DeepSeek-R1-Distill-Qwen-1.5B, #RL Steps=10002026.034,91959.80.42
ThinkPrune-4kBackbone=DeepSeek-R1-Distill-Qwen-1.5B, #RL Steps=N/A2026.034,80260.950.53
Truncated Think 8kBackbone=DeepSeek-R1-Distill-Qwen-7B, #RL Steps=N/A2026.034,12747.2-1.25
Truncated Think 4kBackbone=Qwen3-4B-Thinking-2507, #RL Steps=N/A2026.034,07118.8-3.25
LASER-DE-4096Backbone=DeepSeek-R1-Distill-Qwen-7B, #RL Steps=10002026.033,809710.41
Truncated Think 4kBackbone=DeepSeek-R1-Distill-Qwen-1.5B, #RL Steps=N/A2026.033,62541.9-0.91
Truncated Think 4kBackbone=DeepSeek-R1-Distill-Qwen-7B, #RL Steps=N/A2026.033,58343-1.48
ShorterBetterBackbone=DeepSeek-R1-Distill-Qwen-7B, #RL Steps=2002026.033,49168.30.26
ShorterBetterBackbone=DeepSeek-R1-Distill-Qwen-1.5B, #RL Steps=3002026.032,32752.30.07