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
351Reka Flash 21B20240226· LMArena data1,123.8
352InternLM2.5 20B Chat· LMArena data1,122.7
353DeepSeek Coder V2 Instruct· LMArena data1,120.2
354Granite 3.1 2B Instruct· LMArena data1,115.9
355Zephyr ORPO 141B A35B v0.1· LMArena data1,113.1
356Mixtral 8x7B Instruct v0.1· LMArena data1,108.7
357Yi 34B Chat· LMArena data1,108
358Phi-3 Medium 4K Instruct· LMArena data1,106.4
359Tulu 2 DPO 70B· LMArena data1,104.9
360DBRX Instruct Preview· LMArena data1,104.2
361Gemini Pro· LMArena data1,103.8
362Zephyr 7B Beta· LMArena data1,102.2
363Nous Hermes 2 Mixtral 8x7B DPO· LMArena data1,101.6
364Starling LM 7B Alpha· LMArena data1,099.1
365OpenHermes 2.5 Mistral 7B· LMArena data1,097.1
366Starling LM 7B Beta· LMArena data1,095.6
367Llama 3.2 3B· LMArena data1,094.1
368OpenChat 3.50106· LMArena data1,092.4
369GPT-3.5 Turbo0125· LMArena data1,092.3
370Qwen1.5-14B-Chat· LMArena data1,090.4
371Llama 2 70B SteerLM Chat· LMArena data1,090.4
372WizardLM 13B· LMArena data1,090.4
373SOLAR 10.7B Instruct v1.0· LMArena data1,089.3
374Falcon 180B Chat· LMArena data1,088.5
375Dolphin 2.2.1 Mistral 7B· LMArena data1,085.6
376Qwen1.5-32B-Chat· LMArena data1,082.8
377Phi-3 Small 8K Instruct· LMArena data1,082.7
378Snowflake Arctic Instruct· LMArena data1,080.8
379Guanaco 33B· LMArena data1,079.9
380Llama 2 70B Chat· LMArena data1,074.8
381MPT 30B Chat· LMArena data1,074
382Granite 3.0 8B Instruct· LMArena data1,070.6
383Zephyr 7B Alpha· LMArena data1,068.7
384Mistral 7B Instruct v0.2· LMArena data1,067.6
385DeepSeek LLM 67B Chat· LMArena data1,067.3
386Gemma 1.1 7B IT· LMArena data1,059.3
387Vicuna 13B· LMArena data1,058
388Llama 2 13B Chat· LMArena data1,046.8
389Granite 3.0 2B Instruct· LMArena data1,046.1
390Phi-3 Mini 4K InstructJune 2024· LMArena data1,044.6
391Phi-3 Mini 4K Instruct· LMArena data1,036.8
392GPT-3.5 Turbo1106· LMArena data1,035
393Qwen1.5-7B-Chat· LMArena data1,034.4
394Llama 2 7B Chat· LMArena data1,033.2
395Llama 3.2 1B Instruct· LMArena data1,032.7
396Code Llama 34B Instruct· LMArena data1,031.5
397SmolLM2 1.7B Instruct· LMArena data1,031.3
398StripedHyena Nous 7B· LMArena data1,029.2
399Mistral 7B Instruct· LMArena data1,028.3
400Qwen-14B-Chat· LMArena data1,027.6

Showing 351–400 of 421 results.

Data credits: LMArena, CC BY 4.0 (licence)