Skip to content

Benchmark

Arena ELO - Creative Writing

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

Models measured
364
Scores recorded
421
Sources
1
IQ weight
1.14

Score distribution

1,287.4
Mean
1,318.1
Median
144.2
Std dev
1,079.9
10th percentile
1,442.3
90th percentile
793.9
Min
793.91,532.7
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.14 for chat models, measured over 364 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
301Olmo 3.1 32B Think· LMArena data1,226.5
302Gemini 1.5 Flash001· LMArena data1,222
303Qwen2.5 72B Instruct· LMArena data1,221.2
304Gemini 1.5 Flash-8B001· LMArena data1,217.4
305Llama 3.1 Nemotron 51B Instruct· LMArena data1,213.2
306Nova Pro 1.0· LMArena data1,212.2
307Jamba 1.5 Large· LMArena data1,211.6
308Granite 4.0 H Small· LMArena data1,210.8
309Llama 3 70B Instruct· LMArena data1,209.7
310Gemma 2 9B IT· LMArena data1,205.5
311Granite 4.2 3B· LMArena data1,203.8
312Nemotron-4 340B Instruct· LMArena data1,202.4
313Reka Flash20240904· LMArena data1,202.1
314GPT OSS 20b· LMArena data1,201.7
315Command R+· LMArena data1,200.9
316Aya Expanse 32B· LMArena data1,199.7
317OLMo 2 32B Instruct0325· LMArena data1,198.4
318GLM 40520· LMArena data1,197.9
319Nova Lite 1.0· LMArena data1,196.8
320Mistral Small 24B Instruct2501· LMArena data1,194.6
321GPT 40613· LMArena data1,191.4
322GPT 40314· LMArena data1,190.1
323Mercury· LMArena data1,186.5
324Claude 3 Sonnet20240229· LMArena data1,186.1
325Phi 4· LMArena data1,181.5
326Llama 3.1 Tulu 3 8B· LMArena data1,181.4
327Qwen2 72B Instruct· LMArena data1,180.8
328Ministral 8B2410· LMArena data1,174.7
329Qwen 2.5 Coder 32B Instruct· LMArena data1,173.8
330Nova Micro 1.0· LMArena data1,171.7
331Aya Expanse 8B· LMArena data1,168
332Hunyuan Standard 256K· LMArena data1,165
333Jamba 1.5 Mini· LMArena data1,164.3
334Mistral Large2402· LMArena data1,161
335Mistral Medium· LMArena data1,157.6
336Claude 3 Haiku20240307· LMArena data1,157.2
337Llama 3.1 8B Instruct· LMArena data1,154.2
338Llama 3 8B Instruct· LMArena data1,149.7
339WizardLM 70B· LMArena data1,149.1
340Qwen1.5-110B-Chat· LMArena data1,147.8
341Gemma 2 2B IT· LMArena data1,146.7
342Mixtral 8x22B Instruct v0.1· LMArena data1,141.3
343Qwen1.5-72B-Chat· LMArena data1,137
344Yi 1.5 34B Chat· LMArena data1,135.1
345QwQ 32B Preview· LMArena data1,133.7
346Reka Flash 21BOnline, 20240226· LMArena data1,130.5
347Gemini ProDev API· LMArena data1,130.4
348Granite 3.1 8B Instruct· LMArena data1,128.8
349OpenChat 3.5· LMArena data1,125.5
350Vicuna 33B· LMArena data1,124.6

Showing 301–350 of 421 results.

Data credits: LMArena, CC BY 4.0 (licence)