Skip to content

Benchmark

Arena ELO Text Factuality - Math

Math benchmarks measure how well models solve maths problems. Higher scores are better.

Models measured
107
Scores recorded
122
Sources
1
IQ weight
2.16

Score distribution

1,450.8
Mean
1,446
Median
32.8
Std dev
1,411.7
10th percentile
1,493.9
90th percentile
1,358
Min
1,3581,518.4
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.16 for chat models, measured over 107 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51Inkling Small· LMArena data1,455.2
52Grok 4.20Reasoning, 0309, beta· LMArena data1,455.1
53Kimi K2.5Thinking· LMArena data1,454.5
54GPT 5.2High· LMArena data1,454.5
55DeepSeek-V4-Pro PreviewHigh· LMArena data1,453.2
56GLM 5· LMArena data1,452.8
57Qwen3.6 Plus· LMArena data1,450.9
58Gemini 3 FlashMinimal· LMArena data1,449.9
59GPT-6 LunaMax· LMArena data1,449.5
60Gemini 2.5 Pro· LMArena data1,447.8
61GPT 5.1High· LMArena data1,446
62Claude Sonnet 4.5High, 32K, 20250929· LMArena data1,446
63MiMo V2.5· LMArena data1,444
64Grok 4.7xHigh· LMArena data1,444
65Claude Haiku 5.5High· LMArena data1,443.7
66ERNIE 5.00110· LMArena data1,443.7
67MiMo V2 Pro· LMArena data1,443.4
68Grok 4.20 Multi-Agent0309, beta· LMArena data1,442.9
69GPT-5.2 Chat20260210· LMArena data1,442.8
70GPT-5.4 miniHigh· LMArena data1,442.8
71DeepSeek V4 Pro· LMArena data1,441.2
72Qwen 3.8 27B· LMArena data1,441
73Grok 4.6High· LMArena data1,440.6
74Gemini 3.5 Flash Lite· LMArena data1,440.2
75DeepSeek-V4-Flash PreviewHigh· LMArena data1,439.9
76GLM 4.6· LMArena data1,439.9
77MiMo V2 Omni· LMArena data1,438.9
78GPT-6 SolMax· LMArena data1,437.9
79GPT 5.2· LMArena data1,437.4
80Grok 4.20 Beta 1· LMArena data1,437
81Qwen3 235B A22B Instruct 25072507· LMArena data1,436
82Claude Opus 4.1Thinking 16K, 20250805· LMArena data1,435.6
83Qwen3.5 27B· LMArena data1,435
84Claude Opus 4.120250805· LMArena data1,435
85nova-experimental-chat11-10· LMArena data1,434.7
86Kimi K2 Thinking· LMArena data1,434.3
87Qwen3.5 122B-A10B· LMArena data1,433.9
88GPT 5.1· LMArena data1,432.8
89Gemini 3.1 Flash Lite· LMArena data1,430.8
90DeepSeek V3.2· LMArena data1,430.6
91MiniMax M3· LMArena data1,429.8
92LongCat-Flash-Chat Exp2602· LMArena data1,429.3
93GPT-5.5 Instant· LMArena data1,429.1
94MiniMax M2.7· LMArena data1,427.2
95DeepSeek V4 Flash· LMArena data1,426.2
96Claude Sonnet 4.520250929· LMArena data1,425.6
97Solar Pro 4· LMArena data1,423.9
98DeepSeek V3.2Thinking· LMArena data1,423.4
99Grok 4.1Thinking· LMArena data1,423
100GPT 5.4 NanoHigh· LMArena data1,422.3

Showing 51–100 of 122 results.

Data credits: LMArena, CC BY 4.0 (licence)