Inference
EN
Vals AI

MATH 500

Academic math benchmark on probability, algebra, and trigonometry

ModelsModels with a result.59
Top resultBest result on this benchmark.96.4%Gemini 3 Pro
Top-3 spreadPoints from first to third.0.4 pts
UpdatedDate the source changed the results.9 Jan 2026

Result and cost

57 of 59 results

Results

59 results
#ModelResultThe score from the source.CostUS dollars to run one task.TimeTime to run one task.
01Gemini 3 Pro96.4% ±0.8$0.05132 s
02Grok 496.2% ±0.9$0.03727 s
03GPT-596.0% ±0.9$0.0232 s
04Claude Opus 4.1 Thinking95.4% ±0.9$0.20246 s
05Gemini 2.5 Pro Exp 03.2595.2% ±1.0$0.007926 s
06GPT-5 mini94.8% ±1.0$0.003712 s
06GPT-OSS 120B94.8% ±1.0$0.001416 s
08o394.6% ±1.0$0.01117 s
08Qwen3 235B-A22B94.6% ±1.0$0.00342 min
10o4-mini94.2% ±1.0$0.004813 s
10Kimi K2 Instruct94.2% ±1.0$0.00337 s
10GPT-OSS 20B94.2% ±1.0$0.0009631 s
10Grok 3 Mini Fast High Reasoning94.2% ±1.0$0.001923 s
14GLM-4.594.0% ±1.1$0.01385 s
15Claude Sonnet 4 Thinking93.8% ±1.1$0.06863 s
15GPT-5 nano93.8% ±1.1$0.001722 s
17Claude Opus 4.193.0% ±1.1$0.04631 s
18DeepSeek-R192.2% ±2.4$0.0113 min
19o3-mini91.8% ±2.4$0.007714 s
19Gemini 2.5 Flash Preview 04.17 Thinking91.8% ±1.2$0.01124 s
21Claude 3.7 Sonnet Thinking91.6% ±1.2$0.1252 min
21Gemini 2.5 Flash Preview 04.1791.6% ±1.2$0.00659.5 s
23Llama 3.3 Nemotron Super 49B V1 42e84561 Thinking91.4% ±1.3—43 s
24o190.4% ±2.6$0.09524 s
24Claude Opus 490.4% ±1.3$0.04515 s
26Claude Sonnet 490.3% ±1.3$0.0110 s
27Grok 3 [Beta]89.8% ±1.4$0.0139.1 s
28MiniMax M2.189.0% ±1.4$0.004660 s
28Gemini 2.0 Flash Exp89.0% ±2.7$0.00023.4 s
30DeepSeek V3 032488.6% ±1.4$0.0007413 s
31GPT-4.1 mini88.0% ±1.5$0.00125.8 s
31Gemini 2.0 Flash 00188.0% ±2.9$0.000263.4 s
33GPT-4.187.2% ±1.5$0.009413 s
34Mistral Medium87.0% ±1.5$0.001814 s
35Llama4 Maverick Instruct Basic85.2% ±1.6$0.000597.5 s
36Gemini 2.0 Flash Thinking Exp 01.2184.6% ±3.2$0.0003912 s
37Gemini 1.5 Pro 00282.8% ±3.3$0.00225.0 s
38DeepSeek V380.4% ±1.8$0.000567.9 s
39GPT-4.1 nano80.2% ±1.8$0.000313.4 s
40Llama 4 Scout 17B 16E Instruct79.2% ±1.8$0.0004311 s
41Gemini 1.5 Flash 00278.8% ±3.6$0.000122.6 s
42Grok 278.4% ±3.6$0.006320 s
43Claude 3.7 Sonnet76.8% ±1.9$0.00695.5 s
44Command A76.2% ±1.9$0.0068.7 s
45Mistral Large 241174.4% ±3.8$0.00339.9 s
46GPT-4o74.0% ±3.8$0.008813 s
47Llama 3.3 70B Instruct Turbo73.4% ±3.9$0.00075.4 s
48GPT-4o mini72.6% ±3.9$0.000356.3 s
49Claude 3.5 Sonnet72.4% ±3.9$0.0054.6 s
50Meta Llama 3.1 405B Instruct Turbo71.4% ±3.9$0.00345 s
51Llama 3.3 Nemotron Super 49B V1 42e8456171.2% ±2.0—13 s
52Mistral Small 240270.6% ±4.0$0.000334.9 s
53Grok 3 Mini Fast Low Reasoning70.2% ±2.0$0.000899.9 s
54Mistral Small 3.168.4% ±2.1$0.000166.4 s
55Meta Llama 3.1 70B Instruct Turbo65.0% ±4.2$0.0019.2 s
56Claude 3.5 Haiku64.2% ±4.2$0.00155.1 s
57Jamba Large 1.654.8% ±2.2$0.00413 s
58Meta Llama 3.1 8B Instruct Turbo44.4% ±4.3$0.000355.8 s
59Jamba Mini 1.625.4% ±1.9$0.000314.9 s