Skip to content

Benchmark

Arena ELO - Korean

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

Models measured
244
Scores recorded
288
Sources
1
IQ weight
1.99

Score distribution

1,298.9
Mean
1,348.8
Median
139.4
Std dev
1,087.1
10th percentile
1,441.2
90th percentile
856
Min
8561,509.5
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.99 for chat models, measured over 244 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
201Mistral Small 3.22506· LMArena data1,255.4
202Nemotron 3.5 Lightning 30B A3BNVFP4· LMArena data1,252.4
203Claude Sonnet 3.7Thinking 32K, 20250219· LMArena data1,248.9
204DeepSeek V3· LMArena data1,248.5
205GLM 4 Plus· LMArena data1,243.5
206Claude Sonnet 3.720250219· LMArena data1,238.6
207Grok 22024-08-13· LMArena data1,236.9
208GPT 4o2024-05-13· LMArena data1,233.9
209MiniMax M2.5· LMArena data1,233.7
210GPT OSS 20b· LMArena data1,233.7
211Gemini Advanced0514· LMArena data1,232
212Gemini 1.5 Pro001· LMArena data1,224.5
213O1 Mini· LMArena data1,222.9
214Gemini 1.5 Flash002· LMArena data1,220.5
215GPT 4o2024-08-06· LMArena data1,218.3
216DeepSeek V2.5· LMArena data1,209.5
217Llama 4 Scout· LMArena data1,206.7
218Mistral Small 3.1 24B Instruct2503· LMArena data1,206.4
219Athene V2 Chat· LMArena data1,205.8
220Olmo 3.1 32B Instruct· LMArena data1,203.7
221Llama 4 Maverick· LMArena data1,203.3
222Nova Pro 1.0· LMArena data1,202.8
223Mistral Large2411· LMArena data1,201.7
224Gemini 1.5 Flash001· LMArena data1,201.1
225Claude 3.5 Sonnet20240620· LMArena data1,200.4
226Claude 3.5 Sonnet20241022· LMArena data1,197.5
227GPT-4o mini2024-07-18· LMArena data1,195.2
228Grok 2 mini2024-08-13· LMArena data1,194.9
229Yi Lightning· LMArena data1,192.4
230Mistral Small 24B Instruct2501· LMArena data1,188.2
231Qwen2.5 72B Instruct· LMArena data1,188
232Claude 3 Opus20240229· LMArena data1,187.4
233GPT-4 Turbo2024-04-09· LMArena data1,186.7
234Llama 3.1 405B InstructFP8· LMArena data1,184
235GPT-4 Turbo Preview1106· LMArena data1,183.9
236Gemma 2 27B IT· LMArena data1,173.8
237GPT-4 Turbo Preview0125· LMArena data1,173.6
238Claude Haiku 3.520241022· LMArena data1,173.1
239Llama 3.1 405B InstructBF16· LMArena data1,172.4
240Mistral Large 22407· LMArena data1,170
241Command R+08-2024· LMArena data1,162.5
242Aya Expanse 32B· LMArena data1,157.9
243Nova Lite 1.0· LMArena data1,153.1
244Phi 4· LMArena data1,150.9
245Athene 70B0725· LMArena data1,150.7
246Nova Micro 1.0· LMArena data1,150.3
247Nemotron-4 340B Instruct· LMArena data1,147.9
248Llama 3.3 70B Instruct· LMArena data1,142.5
249Qwen Max0919· LMArena data1,141.6
250Llama 3.1 70B Instruct· LMArena data1,140.5

Showing 201–250 of 288 results.

Data credits: LMArena, CC BY 4.0 (licence)