Skip to content

Benchmark

Arena ELO - English

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

Models measured
366
Scores recorded
423
Sources
1
IQ weight
1.26

Score distribution

1,327.7
Mean
1,373.1
Median
138.5
Std dev
1,119.5
10th percentile
1,468.7
90th percentile
845.9
Min
845.91,532
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.26 for chat models, measured over 366 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
301Magistral Medium2506· LMArena data1,274.1
302Llama 3.1 Tulu 3 70B· LMArena data1,272.7
303Claude Haiku 3.520241022· LMArena data1,267.6
304Llama 3.1 Nemotron 51B Instruct· LMArena data1,267.1
305Jamba 1.5 Large· LMArena data1,266.7
306Granite 4.0 H Small· LMArena data1,266.5
307Claude 3 Opus20240229· LMArena data1,262.8
308Mistral Small 24B Instruct2501· LMArena data1,256.1
309Reka Core20240904· LMArena data1,256
310OLMo 2 32B Instruct0325· LMArena data1,255.4
311Qwen 2.5 Coder 32B Instruct· LMArena data1,247.9
312Gemini 1.5 Flash001· LMArena data1,243.7
313Nova Lite 1.0· LMArena data1,242.6
314Gemma 2 9B IT SimPO· LMArena data1,242.4
315GLM 40520· LMArena data1,242
316Gemma 2 27B IT· LMArena data1,241.9
317Command R+08-2024· LMArena data1,237.6
318Nemotron-4 340B Instruct· LMArena data1,236.5
319Gemini 1.5 Flash-8B001· LMArena data1,231.3
320Phi 4· LMArena data1,231
321Aya Expanse 32B· LMArena data1,228.9
322Claude 3 Sonnet20240229· LMArena data1,226.1
323Jamba 1.5 Mini· LMArena data1,224
324Nova Micro 1.0· LMArena data1,223.8
325Reka Flash20240904· LMArena data1,222.8
326Qwen2 72B Instruct· LMArena data1,222.1
327GPT 40314· LMArena data1,220.1
328Gemma 2 9B IT· LMArena data1,219.7
329Llama 3.1 8B Instruct· LMArena data1,217.5
330Hunyuan Standard 256K· LMArena data1,214.9
331Llama 3.1 Tulu 3 8B· LMArena data1,214.9
332Yi 1.5 34B Chat· LMArena data1,214.8
333Llama 3 8B Instruct· LMArena data1,213.3
334Command R+· LMArena data1,213.2
335Ministral 8B2410· LMArena data1,212.5
336Claude 3 Haiku20240307· LMArena data1,206.1
337GPT 40613· LMArena data1,205.3
338InternLM2.5 20B Chat· LMArena data1,203
339Mistral Large2402· LMArena data1,199.9
340Qwen1.5-110B-Chat· LMArena data1,198
341DeepSeek Coder V2 Instruct· LMArena data1,194.5
342Aya Expanse 8B· LMArena data1,190.8
343Mistral Medium· LMArena data1,190.3
344Granite 3.1 8B Instruct· LMArena data1,189.2
345Mixtral 8x22B Instruct v0.1· LMArena data1,187.2
346Qwen1.5-72B-Chat· LMArena data1,186.5
347QwQ 32B Preview· LMArena data1,184.3
348Gemma 2 2B IT· LMArena data1,181.5
349Reka Flash 21BOnline, 20240226· LMArena data1,180.8
350Llama 3.2 3B· LMArena data1,179.6

Showing 301–350 of 423 results.

Data credits: LMArena, CC BY 4.0 (licence)