Benchmark By Score Rank
Overall 1
AA Intelligence Index4.3.2 Artificial Analysis combines tests of knowledge, reasoning, coding and tool use into one score AA 22.4 #103 of 336
Knowledge 21
Arena TextEnglish Human pairwise preferences in the stated text vote pool Arena 1,475 #23 of 205+13 versions & slices instruction following Human pairwise preferences in the stated text vote pool Arena 1,466 #26 of 205non-English Human pairwise preferences in the stated text vote pool Arena 1,458 #23 of 205Russian Human pairwise preferences in the stated text vote pool Arena 1,467 #25 of 203Chinese Human pairwise preferences in the stated text vote pool Arena 1,522 #21 of 198overall · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,470 #23 of 198overall Human pairwise preferences in the stated text vote pool Arena 1,476 #24 of 196overall · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,470 #23 of 188Spanish Human pairwise preferences in the stated text vote pool Arena 1,478 #12 of 168German Human pairwise preferences in the stated text vote pool Arena 1,471 #20 of 167French Human pairwise preferences in the stated text vote pool Arena 1,478 #27 of 164Korean Human pairwise preferences in the stated text vote pool Arena 1,438 #19 of 156Japanese Human pairwise preferences in the stated text vote pool Arena 1,449 #22 of 152Polish Human pairwise preferences in the stated text vote pool Arena 1,457 #25 of 142
AA Omniscience Index Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty AA -6.6 #86 of 320
SimpleQA Verifiedv1 Accuracy on short factual questions with verified answers, answered without web access 34.2% #51 of 84
AA OmniscienceAccuracy Measures correct answers on the AA Omniscience factual knowledge test AA 20.3% #155 of 320+1 versions & slices Hallucination Rate Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test AA 33.7% #47 of 320
IFBench (instructions) Checks whether the model follows precise formatting and content instructions AA 73.3% #30 of 245
Mystery Game Puzzles Models choose the best move in puzzle positions from an undisclosed game Epoch 19% #39 of 65
EQ-Bench3 Judge-rated responses to social and emotionally nuanced scenarios, scored through the source pairwise ranking EQ-Bench 1,563 #7 of 53
Reasoning 5
ARC-AGI-2semi-private Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 22.8% #42 of 69+1 versions & slices public evaluation Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 20.8% #39 of 66
Chess Puzzles Models choose the best next move from a novel chess position Epoch 22% #48 of 69
ARC-AGI-1semi-private Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 77% #40 of 67+1 versions & slices public evaluation Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 80.4% #38 of 64
Math 17
FrontierMath Tiers 1–3 (v2) Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 59.2% #36 of 70
FrontierMath Tier 4 (v2) Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 29.3% #30 of 55
MathArenaApex Answer accuracy on the named mathematics competition under the recorded effort MathArena 19.8% #13 of 34+14 versions & slices Apex Shortlist Answer accuracy on the named mathematics competition under the recorded effort MathArena 68.1% #17 of 30AIME 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 90% #22 of 24HMMT February 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 92.4% #12 of 24ArXivMath · June 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 46.5% #21 of 24BrokenArXiv · June 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 14.2% #22 of 24ArXivMath · February 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 46.1% #7 of 20ArXivMath · May 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 56.7% #13 of 19BrokenArXiv · May 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 10.5% #15 of 18ArXivMath · April 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 47.5% #13 of 17BrokenArXiv · April 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 18.4% #12 of 15BrokenArXiv · February 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 10.5% #11 of 15ArXivMath · March 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 61.7% #6 of 14BrokenArXiv · March 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 12.1% #8 of 13ArXivLean · June 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 8.7% #8 of 9
Science 2
Humanity's Last Exam Very hard expert questions across many fields, written to stump today’s models AA 9.8% #197 of 336
CritPt (physics) Research-level physics problems that need many careful reasoning steps AA 3.1% #93 of 323
Health & law 5
MLCR-AA Models synthesize long medical records; the headline credits concise answers judged both accurate and complete AA 12.8% #43 of 103
Vals LegalBench Legal reasoning and legal-language understanding on LegalBench Vals 84.1% #35 of 89
Vals MedScribe Accuracy of clinical notes produced from consultation transcripts Vals 83.5% #40 of 80
Vals MedCode Medical coding accuracy on patient encounters Vals 40.8% #49 of 78
Vals Legal Research Bench Grounded legal research; a task passes only if all criteria pass Vals 31.3% #41 of 73
Documents 2
AA-LCR (long context)1.1 Questions that need reading and connecting facts across very long documents (around 100K tokens) AA 42.3% #191 of 314
GDP.pdfAA all-pass Models reason over professional PDF documents; an attempt passes only when every atomic criterion passes 10.4% #60 of 135
Legacy · saturated or no longer reported 3
GPQA Diamond Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 68.6% #191 of 312+1 versions & slices Epoch Models answer graduate-level science multiple-choice questions Epoch 91.9% #20 of 134
OTIS Mock AIME 20242025 Models solve the stated OTIS mock AIME mathematics problems Epoch 86.4% #62 of 131
No benchmark matches
CheaperDeepSeek V4.1 Flash 1017maxOpensame-76%$0.30 / $1.20Memory (4-bit): 452 GB → 458 GB+5
BetterGPT-5.5xhighwins 58%+423%$2.10Prompts with >272K input tokens
CheaperStep 3.7 FlashunstatedOpensame-80%$0.20 / $1.15Context: 1M → 256K
Kimi K2.6unstatedOpensame-20%$0.95 / $4Context: 1M → 256K+1
Claude Opus 4.8maxwins 54%+365%$4.08