Skip to content

Benchmark

Arena ELO - Expert

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

Models measured
318
Scores recorded
371
Sources
1
IQ weight
1.30

Score distribution

1,333.9
Mean
1,371.6
Median
146.5
Std dev
1,113
10th percentile
1,491.8
90th percentile
953.7
Min
953.71,581.3
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.30 for chat models, measured over 318 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
101Inkling Small· LMArena data1,440.7
102nova-experimental-chat26-01-10· LMArena data1,440.6
103GPT 5.1· LMArena data1,440.5
104Doubao Seed 2.0 Pro· LMArena data1,439.8
105Kimi K2.5Instant· LMArena data1,439.7
106Kimi K2 Thinking· LMArena data1,439.1
107Claude Opus 4.1Thinking 16K, 20250805· LMArena data1,439
108GPT-6 SolMax· LMArena data1,439
109Grok 4.20Reasoning, 0309, beta· LMArena data1,438.9
110Claude Haiku 5.5High· LMArena data1,438
111GPT-5.2 Chat20260210· LMArena data1,437.6
112GPT-6 LunaMax· LMArena data1,436.7
113DeepSeek-V4-Flash PreviewHigh· LMArena data1,436.4
114GPT-5.4 miniHigh· LMArena data1,435.9
115DeepSeek V3.2· LMArena data1,435
116GLM 4.5· LMArena data1,433
117Gemini 3.5 Flash Lite· LMArena data1,432.8
118GLM 4.6· LMArena data1,432.8
119DeepSeek V3.2Thinking· LMArena data1,432.3
120Qwen3.5 122B-A10B· LMArena data1,431.7
121Mistral Medium 3.5· LMArena data1,431.6
122MiniMax-M2.1 Preview· LMArena data1,431.5
123Grok 4.7xHigh· LMArena data1,430.8
124nova-experimental-chat11-10· LMArena data1,429.3
125Qwen3 VL 235Bthinking· LMArena data1,429
126Qwen3.5 27B· LMArena data1,428.7
127Grok 4.20 Beta 1· LMArena data1,428.4
128GPT 5.2· LMArena data1,428.2
129ERNIE 5.00110· LMArena data1,426.6
130Gemini 2.5 Flash· LMArena data1,426.5
131Claude Opus 4.120250805· LMArena data1,425.7
132MiMo V2 FlashNon-Thinking· LMArena data1,425.4
133Solar Pro 4· LMArena data1,424.7
134GLM 4.7· LMArena data1,423.7
135Mistral Large 3· LMArena data1,422.4
136ERNIE-5.0-Preview1203· LMArena data1,422.3
137DeepSeek V3.2 ExpThinking· LMArena data1,422.1
138NVIDIA Nemotron 3.5 Super VL 120BNVFP4· LMArena data1,421.9
139Grok 3 Preview02-24· LMArena data1,421.2
140Step 3.5 Flash· LMArena data1,420.4
141Gemini 3 FlashMinimal· LMArena data1,420.3
142GPT 5High· LMArena data1,419.5
143Gemini 2.5 Flash Preview09-2025· LMArena data1,419.1
144ERNIE-5.0-Preview1022· LMArena data1,418
145Qwen3 Next 80B A3B Instruct· LMArena data1,417.3
146Grok 4.1Thinking· LMArena data1,417.1
147Grok 40709· LMArena data1,415.2
148Grok 4 FastReasoning· LMArena data1,413
149Grok 4.1· LMArena data1,411.8
150Granite 4.2 30B· LMArena data1,409.2

Showing 101–150 of 371 results.

Data credits: LMArena, CC BY 4.0 (licence)