Skip to content

Benchmark

MATH

Math benchmarks measure how well models solve maths problems. Scores are percentages. Higher is better.

Models and systems measured
69
Scores recorded
69
Sources
2
IQ weight
1.37

Score distribution

68.5%
Mean
71.8%
Median
17%
Std dev
43.1%
10th percentile
87.6%
90th percentile
12.4%
Min
12.4%97.9%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.37 for chat models, measured over 61 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1o3 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim97.9%
2o1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim96.4%
3Ministral 3 14B Instruct2512· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim90.4%
4Mistral Large 3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim90.4%
5Gemini 2.0 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim89.7%
6Kimi K2 0905· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim89.1%
7Gemma 3 27B IT· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim89%
8Ministral 3 8b Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim87.6%
9Gemini 2.0 Flash Lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim86.8%
10Gemini 1.5 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim86.5%
11MiMo V2.5 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim86.2%
12O1 Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim85.5%
13GPT 5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim84.7%
14Gemma 3 12B IT· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim83.8%
15Qwen2.5 32B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim83.1%
16Qwen2.5 72B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim83.1%
17Ministral 3 3b Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim83%
18Phi 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim80.4%
19Qwen 2.5 14B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim80%
20Gemini 1.5 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim77.9%
21Llama 3.3 70B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim77%
22GPT 4o· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim76.6%
23Nova Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim76.6%
24Grok 2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim76.1%
25Gemma 3 4B IT· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim75.6%
26Qwen2.5 7B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim75.5%
27DeepSeek V2.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim74.7%
28Llama 3.1 405B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim73.8%
29Nova Lite· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim73.3%
30Grok 2 mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim73%
31GPT-4 Turbo· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim72.6%
32Qwen 3 235B A22B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim71.8%
33Qwen2.5-Omni 7B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim71.5%
34Claude 3.5 Sonnet· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim71.1%
35Mistral Small 24B Instruct2501· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim70.6%
36GPT-4o mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim70.2%
37Mistral Small 3.2 24B Instruct2506· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim69.4%
38Claude Haiku 3.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim69.4%
39Mistral Small 3.1 24b Instruct 2503· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim69.3%
40Nova Micro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim69.3%
41Llama 3.1 70B Instruct0-shot · chain of thought· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.68%
42Phi 4 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim64%
43Llama 4 Maverick· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim61.2%
44Claude 3 Opus· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim60.1%
45Qwen2 72B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim59.7%
46phi-3.5-moe-instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim59.5%
47Gemini 1.5 Flash-8B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim58.7%
48Qwen 2.5 Coder 32B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim57.2%
49Llama 3.2 11B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim51.9%
50Llama 4 Scout· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim50.3%

Showing 1–50 of 63 results.

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
DeepSeek V4.1 Flash BaseEM · 4-shot· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.61.1%
MiMo V2 Flash Baseofficial Base table · 4-shot· Reported by Xiaomi MiMo· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.71%
Qwen3.5-397B-A17B-BaseQwen3.5 release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.74.1%
Qwen3.7-Plus-BaseQwen3.8-Flash-Next release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.74.4%
Qwen3.8-27B-BaseQwen3.8-Flash-Next release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.60.5%
Qwen3.8-Flash-Next-BaseQwen3.8-Flash-Next release/model card · base-model table· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.72.8%

Data credits: Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.