Skip to content

Benchmark

Arena ELO Vision - Entity Recognition

People compare model answers and choose which they prefer. Higher scores are better.

Models measured
44
Scores recorded
51
Sources
1
IQ weight
1.50

Score distribution

1,234.7
Mean
1,228.7
Median
48.5
Std dev
1,183.1
10th percentile
1,288.8
90th percentile
1,140.6
Min
1,140.61,346.5
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.50 for chat models, measured over 44 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 3 Pro Preview· LMArena data1,346.5
2Gemini 3 Flash· LMArena data1,335.1
3Gemini 3.1 Pro Preview· LMArena data1,327.3
4Gemini 3 FlashMinimal· LMArena data1,311.1
5Gemini 3.1 Flash Lite· LMArena data1,306.4
6Grok 40709· LMArena data1,288.8
7Kimi K2.5Thinking· LMArena data1,286.6
8Gemini 2.5 Pro· LMArena data1,282.7
9Claude Opus 4.7High· LMArena data1,278.8
10Grok 4.20Reasoning, 0309, beta· LMArena data1,277.6
11GLM 5.3 Flash· LMArena data1,274.9
12Claude Opus 4.6· LMArena data1,272.9
13Claude Opus 4.6High· LMArena data1,270.6
14Qwen3.5 397B A17B· LMArena data1,269.7
15Grok 4.20 Multi-Agent0309, beta· LMArena data1,263.1
16GPT 5High· LMArena data1,261.7
17Grok 4.1 FastReasoning· LMArena data1,254.2
18Gemini 2.5 Flash· LMArena data1,252.5
19Qwen3 VL 235B· LMArena data1,248.7
20GPT 5.4· LMArena data1,246.4
21GPT-5.2 Chat20260210· LMArena data1,245.8
22ChatGPT-4o Latest20250326· LMArena data1,245.5
23OpenAI o32025-04-16· LMArena data1,234
24GPT 5.2High· LMArena data1,232.2
25GPT 5.1High· LMArena data1,231.9
26GPT 5.5· LMArena data1,228.7
27GPT 5.5High· LMArena data1,228.5
28Gemma 4 31B IT· LMArena data1,227.8
29GPT-5 miniHigh· LMArena data1,226.2
30GPT 5.4High· LMArena data1,225.7
31Claude Opus 4.7· LMArena data1,224.5
32GPT-5.4 miniHigh· LMArena data1,224.3
33o4 Mini2025-04-16· LMArena data1,220.5
34Grok 4.3· LMArena data1,219.7
35GPT-5 Chat· LMArena data1,217.5
36Qwen3.5 27B· LMArena data1,212.3
37GPT 4.12025-04-14· LMArena data1,208.5
38GPT 5.2· LMArena data1,198.7
39Gemma 3 27B IT· LMArena data1,197.3
40Claude Sonnet 4.6· LMArena data1,195.7
41MiMo V2.5· LMArena data1,194.5
42MiniMax M3· LMArena data1,194.3
43Gemma 4 26B A4B· LMArena data1,189
44GPT-4.1 mini2025-04-14· LMArena data1,187.1
45Gemini 2.5 Flash-Lite PreviewThinking, 06-17· LMArena data1,186.3
46GPT 5.1· LMArena data1,183.1
47Mistral Small 3.1 24B Instruct2503· LMArena data1,155
48Mistral Small 3.22506· LMArena data1,151.4
49Mistral Medium 3.12508· LMArena data1,147.9
50GPT 5.4 NanoHigh· LMArena data1,141.5

Showing 1–50 of 51 results.

Data credits: LMArena, CC BY 4.0 (licence)