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 6.7 #286 of 339
Knowledge 3
IFBench (instructions) Checks whether the model follows precise formatting and content instructions AA 31% #213 of 248
SimpleQA Verifiedv1 Accuracy on short factual questions with verified answers, answered without web access 9.1% #80 of 84
Mystery Game Puzzles Models choose the best move in puzzle positions from an undisclosed game Epoch 12% #52 of 65
Reasoning 3
ARC-AGI-2public evaluation Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 0% #58 of 66+1 versions & slices semi-private Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 0% #62 of 69
Chess Puzzles Models choose the best next move from a novel chess position Epoch 3% #67 of 69
Math 2
FrontierMath Tiers 1–3 (v2) Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 0.7% #69 of 70
MATH Level5 Models solve competition mathematics problems from the hardest MATH difficulty level Epoch 52.6% #20 of 35
Science 1
Humanity's Last Exam Very hard expert questions across many fields, written to stump today’s models AA 4.2% #306 of 339
Agents 2
GDPval-AA (work tasks)2.1 Real work tasks from many professions (reports, spreadsheets, analyses), graded against expert work AA 0 #146 of 177
τ³-BankingAA Agents retrieve banking knowledge and execute tools to resolve customer-support workflows AA 2.9% #132 of 134
Multimodal 5
MMMU-Pro (vision) College-level questions that require understanding images: charts, diagrams, photos AA 41.5% #122 of 128
MMMUvalidation Image-grounded university-level questions across 30 disciplines; validation and test sets remain separate MMMU 59.4% #25 of 41
OCRBenchv2 · CN Tests visual text recognition, localization and reasoning; English and Chinese subsets remain separate OCRBench 25% #4 of 4+1 versions & slices v2 · EN Tests visual text recognition, localization and reasoning; English and Chinese subsets remain separate OCRBench 43% #3 of 4
Arena Vision Human pairwise preferences in the stated vision vote pool Arena 1,066 #81 of 88
Writing 18
Arena Textcreative writing · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,268 #145 of 197+16 versions & slices creative writing Human pairwise preferences in the stated text vote pool Arena 1,294 #133 of 195overall Human pairwise preferences in the stated text vote pool Arena 1,318 #145 of 196Chinese Human pairwise preferences in the stated text vote pool Arena 1,265 #157 of 198English Human pairwise preferences in the stated text vote pool Arena 1,301 #163 of 205French Human pairwise preferences in the stated text vote pool Arena 1,297 #133 of 164German Human pairwise preferences in the stated text vote pool Arena 1,272 #127 of 167instruction following Human pairwise preferences in the stated text vote pool Arena 1,258 #160 of 205Japanese Human pairwise preferences in the stated text vote pool Arena 1,216 #118 of 152Korean Human pairwise preferences in the stated text vote pool Arena 1,195 #131 of 156non-English Human pairwise preferences in the stated text vote pool Arena 1,266 #158 of 205Polish Human pairwise preferences in the stated text vote pool Arena 1,244 #131 of 142Russian Human pairwise preferences in the stated text vote pool Arena 1,275 #153 of 203Spanish Human pairwise preferences in the stated text vote pool Arena 1,276 #142 of 168overall · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,286 #148 of 188creative writing · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,268 #138 of 187overall · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,286 #154 of 198
EQ-Bench Creative Writingv3 Creative Writing v3 · Elo: Judge-rated creative writing under the v3 prompt set and pairwise ranking EQ-Bench 871 #82 of 94
Health & law 1
MLCR-AA Models synthesize long medical records; the headline credits concise answers judged both accurate and complete AA 0% #101 of 103
Legacy · saturated or no longer reported 9
Terminal-Bench2.1 Older, easier version of Terminal-Bench; Still widely reported AA 5.6% #135 of 156
AIME2025 · AA archived AA evaluates the 30 integer-answer problems from the 2025 AIME mathematics competition AA 14.7% #119 of 140
LiveCodeBenchAA archived AA evaluates solutions to newly collected competitive-programming problems AA 23.4% #133 of 153+1 versions & slices official artifact (through April 2025) Mean pass@1 over each model’s available problems in the current official public generation artifact; its latest problem is dated April 7, 2025 LiveCodeBench 35.5% #10 of 10
GPQA Diamond Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 42.6% #275 of 315+1 versions & slices Epoch Models answer graduate-level science multiple-choice questions Epoch 37.7% #118 of 134
Aider Polyglottwo attempts Models edit code for 225 Exercism exercises in six languages, with a second attempt after test feedback Aider 3.6% #15 of 15
MMLU-Proofficial board Graduate-level multiple-choice knowledge and reasoning across 14 subjects, including science, mathematics, law and health 63.1% #6 of 32
OTIS Mock AIME 20242025 Models solve the stated OTIS mock AIME mathematics problems Epoch 6.9% #113 of 131
No benchmark matches
OpenR1 0528unstated×2.4+165%$0.39 / $1.63No image input
BetterGemini 2.5 Pro Preview (May' 25)unstated×2.3——No image input
BetterOpenReasoning Nemotron 32BOpen×2.3+1,134%$2.72 / $4.80No image input+1
BetterEXAONE 4.0 32BunstatedOpen×2.3——No image input
Qwen3 235B A22BunstatedOpen×2.3+367%$0.70 / $2.80Context: 128K → 33K+1