Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Mathematical Reasoning on AIME 2025 (Pass@1, Timing, and Steps)
Loading...
67.9
Pass@1
PS-ada
56.668
59.584
62.5
65.416
May 6, 2026
Pass@1
Eval E2E Time (h)
Train E2E Time (h)
Eval Steps
Train Steps
Avg Step Time (s)
Validation Accuracy
Updated 2mo ago
Evaluation Results
Method
Method
Links
Pass@1
Eval E2E Time (h)
Train E2E Time (h)
Eval Steps
Train Steps
Avg Step Time (s)
Validation Accuracy
PS-ada
Model=Qwen3-8B
2026.05
67.9
36.9
31.3
190
161
699
77.7
PS-ada
Model=Qwen3-4B
2026.05
66.2
29.5
23.1
140
110
757
80.2
Baseline
Model=Qwen3-4B
2026.05
59
49.5
45.4
230
211
775
66
Baseline
Model=Qwen3-8B
2026.05
57.1
45.4
43.8
260
251
628
61
Feedback
Search any
task
Search any
task
