Q1 Which AI gives the answers people prefer most? Arena Text · Bottom 24%#84 Q1 Which AI gives the answers people prefer most? Arena Text · Bottom 24%#84 Q2 Does it have broad knowledge across expert fields? MMLU-Pro · Top 18%#5 Q3 How high was its intelligence index on the old v4 edition? AA Intelligence Index v4 · Bottom 24%#59
Q1 Bottom 24%#84 of 108 1426 Q2 87.3% Q3 20.6 About Arena Text More about Arena Text
Q1 Can it handle customer service by the rules? TAU2 · Bottom 30%#57 Q1 Can it handle customer service by the rules? TAU2 · Bottom 30%#57 Q2 Can it finish long tasks that take people hours? METR Time Horizon · Bottom 30%#8
Q1 Bottom 30%#57 of 79 73.4% Q2 1h 40m About TAU2 More about TAU2
Q1 Can it pull scattered facts from several long documents into one answer? Long Context Reasoning · Bottom 24%#93 Q1 Can it pull scattered facts from several long documents into one answer? Long Context Reasoning · Bottom 24%#93 Q2 Which AI gives the business answers people prefer most? Arena Business, Management & Finance · Bottom 16%#92 Q3 Can it read long stories and read between the lines? Fiction.LiveBench · Top 50%#5 Q4 Can it find company filings and do financial analysis? Finance Agent v1 · Bottom 47%#9
Q1 Bottom 24%#93 of 121 69.3% Q2 1409 Q3 61.1% Q4 43.6% About Long Context Reasoning More about Long Context Reasoning
Q1 Can it carry out multi-step web research? Deep Research Bench · Bottom 42%#11 Q1 Can it carry out multi-step web research? Deep Research Bench · Bottom 42%#11 Q2 Which AI gives the search answers people prefer most? Arena Search · Bottom 4%#29 Q3 Does it avoid making things up when summarizing? Vectara Hallucination Rate · Bottom 28%#30 Q1 Bottom 42%#11 of 17 46.8% Reasoning effort: Budget 2K Q2 1127 Q3 12.0% Q4 88.0% Q5 60.6 About Deep Research Bench More about Deep Research Bench
Q1 Which AI do people prefer most for coding answers? Arena Coding · Bottom 38%#69 Q1 Which AI do people prefer most for coding answers? Arena Coding · Bottom 38%#69 Q2 Can it operate a computer itself to crack the toughest development jobs? TerminalBench Hard · Bottom 37%#51 Q3 How high is its coding index across terminal tasks and scientific code? AA Coding Index · Bottom 23%#78 Q1 Bottom 38%#69 of 108 1498 Q2 31.1% Q3 34.0 Q4 63.6% Q5 39.8% Q6 70.7% Q7 43.7% Q8 6.9% Q9 72.0% Q10 1468 Q11 1.33× About Arena Coding More about Arena Coding
Q1 Which AI gives the writing and literature answers people prefer most? Arena Writing, Literature & Language · Bottom 41%#65 Q1 Which AI gives the writing and literature answers people prefer most? Arena Writing, Literature & Language · Bottom 41%#65 Q2 Which AI do people prefer most in long conversations? Arena Multi-turn · Bottom 30%#77 Q3 Does it follow tricky rules like format and length without missing any? IFBench · Bottom 21%#65 Q1 Bottom 41%#65 of 108 1432 Q2 1439 Q3 53.7% Q4 1433 Q5 1444 Q6 8.36 Q7 1422 About Arena Writing, Literature & Language More about Arena Writing, Literature & Language
Q1 Can it solve problems that stump experts across fields? Humanity's Last Exam · Bottom 16%#104 Q1 Can it solve problems that stump experts across fields? Humanity's Last Exam · Bottom 16%#104 Q2 Can it solve research-level physics problems? CritPt · Bottom 12%#78 Q3 Which AI gives the math answers people prefer most? Arena Math · Bottom 20%#85 Q1 Bottom 16%#104 of 122 12.3% Q2 0.3% Q3 1421 Q4 79.6% Q5 64.4% Q6 73.3 Q7 73.3% Q8 1438 Q9 1426 Q10 98.2% Q11 85.0% Q12 4.5% Q13 4.2% Q14 35.7% Q15 8.6% Q16 58.8% Q17 5.6% Q18 81.6% Q19 37.4 By effort None 6.2% Thinking 12.3%
About Humanity's Last Exam More about Humanity's Last Exam
Q1 Which AI gives the photo and chart answers people prefer most? Arena Vision · Bottom 8%#62 Q1 Which AI gives the photo and chart answers people prefer most? Arena Vision · Bottom 8%#62 Q2 Can it predict how objects will move just from a picture? VPCT · Bottom 50%#6 Q3 Can it tell where a photo was taken from a single image? GeoBench · Bottom 10%#20
Q1 Bottom 8%#62 of 66 1192 Q2 38.0% Q3 49.0% About Arena Vision More about Arena Vision
Q1 Which AI do people prefer most when asking in Korean? Arena Korean · Bottom 28%#67 Q1 Which AI do people prefer most when asking in Korean? Arena Korean · Bottom 28%#67 Q2 Which AI do people prefer most when asking in Japanese? Arena Japanese · Bottom 27%#66 Q3 Which AI do people prefer most on tough questions? Arena Hard Prompts · Bottom 30%#77 Q1 Bottom 28%#67 of 91 1385 Q2 1387 Q3 1457 Q4 1450 Q5 1441 Q6 1433 Q7 38.0% Q8 88.0% Q9 73.6% Q10 79.6% Q11 67.6% Q12 87.0% Q13 84.2% Q14 94.2% Q15 78.9% Q16 86.0% Q17 84.0% Q18 66.7% Q19 35.0% Q20 13.0% Q21 79.0% Q22 87.9% Q23 96.0% Q24 72.0% Q25 73.0% Q26 95.0% Q27 71.0% Q28 78.0% Q29 88.8% Q30 73.0% Q31 6.0% Q32 25.1% About Arena Korean More about Arena Korean
Q Does it call out a nonsense question instead of playing along? BullshitBench v2 · Bottom 41%#86
Q Does it call out a nonsense question instead of playing along? BullshitBench v2 · Bottom 41%#86
Bottom 41%#86 of 143 34.0% About BullshitBench v2 More about BullshitBench v2