Benchmark By Score Rank
Overall 1
AA Intelligence Index4.3.2 AA Intelligence Index 4.3.2: Artificial Analysis combines tests of knowledge, reasoning, coding and tool use into one score AA 20.2 #120 of 337
Knowledge 6
AA OmniscienceAccuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test AA 38.6% #51 of 321+1 versions & slices Hallucination Rate AA Omniscience · Hallucination Rate: Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test AA 88.1% #231 of 321
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty AA -15.6 #109 of 321
IFBench (instructions) IFBench (instructions): Checks whether the model follows precise formatting and content instructions AA 71.4% #43 of 248
EQ-Bench3 EQ-Bench 3: Judge-rated responses to social and emotionally nuanced scenarios, scored through the source pairwise ranking EQ-Bench 1,500 #21 of 50
SimpleQA Verifiedv1 SimpleQA Verified v1: Accuracy on short factual questions with verified answers, answered without web access 52.3% #24 of 83
Reasoning 4
ARC-AGI-1public evaluation ARC-AGI-1 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 64.3% #44 of 61+1 versions & slices semi-private ARC-AGI-1 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 60.8% #46 of 64
ARC-AGI-2public evaluation ARC-AGI-2 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 4.5% #50 of 63+1 versions & slices semi-private ARC-AGI-2 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 6.5% #47 of 66
Math 6
MathArenaAIME 2025 MathArena · aime · AIME 2025: Answer accuracy on the named mathematics competition under the recorded effort MathArena 89.2% #12 of 30+5 versions & slices BRUMO 2025 MathArena · brumo · BRUMO 2025: Answer accuracy on the named mathematics competition under the recorded effort MathArena 95.8% #6 of 24CMIMC 2025 MathArena · cmimc · CMIMC 2025: Answer accuracy on the named mathematics competition under the recorded effort MathArena 82.5% #11 of 22HMMT February 2025 MathArena · hmmt · HMMT Feb 2025: Answer accuracy on the named mathematics competition under the recorded effort MathArena 77.5% #14 of 30IMO 2025 MathArena · imo · IMO 2025: Answer accuracy on the named mathematics competition under the recorded effort MathArena 16.7% #3 of 4SMT 2025 MathArena · smt · SMT 2025: Answer accuracy on the named mathematics competition under the recorded effort MathArena 87.7% #10 of 24
Science 2
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models AA 20.1% #123 of 337
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps AA 1.1% #120 of 324
Coding 1
Terminal-Bench Hard Terminal-Bench Hard (version unstated): Agents solve difficult software, system-administration and data-processing tasks in a terminal AA 37.1% #38 of 242
Multimodal 3
MMMU-Pro (vision) MMMU-Pro (vision): College-level questions that require understanding images: charts, diagrams, photos AA 70.1% #69 of 127
MMMUvalidation MMMU · validation: Image-grounded university-level questions across 30 disciplines; validation and test sets remain separate MMMU 82.9% #2 of 41
Arena Vision Arena Vision · overall: Human pairwise preferences in the stated vision vote pool Arena 1,214 #60 of 88
Writing 4
Arena Textcreative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool Arena 1,383 #81 of 195+1 versions & slices overall Arena Text · overall: Human pairwise preferences in the stated text vote pool Arena 1,432 #74 of 196
EQ-Bench Longform Creative Writing1.11 Longform Creative Writing 1.11: Models plan, revise and write a story across eight turns, graded with a writing rubric EQ-Bench 62.7% #49 of 95
EQ-Bench Creative Writingv3 Creative Writing v3 · Elo: Judge-rated creative writing under the v3 prompt set and pairwise ranking EQ-Bench 1,676 #34 of 91
Health & law 3
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench Vals 83.8% #42 of 89
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters Vals 47.3% #28 of 78
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts Vals 76.7% #57 of 80
Documents 1
AA-LCR (long context)1.1 AA-LCR (long context) 1.1: Questions that need reading and connecting facts across very long documents (around 100K tokens) AA 74.7% #90 of 275
Legacy · saturated or no longer reported 8
τ²-benchAA τ²-Bench: Tests service agents that use tools while coordinating with a simulated user AA 80.7% #78 of 245+2 versions & slices airline · 0.1.3 τ-bench official · 2-airline-0.1.3: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools 52% #7 of 8retail · 0.1.3 τ-bench official · 2-retail-0.1.3: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools 73.9% #5 of 8
AIME2025 · AA archived AIME 2025 · AA archived: AA evaluates the 30 integer-answer problems from the 2025 AIME mathematics competition AA 88.3% #22 of 140
LiveCodeBenchAA archived LiveCodeBench · AA archived: AA evaluates solutions to newly collected competitive-programming problems AA 80.8% #14 of 153+1 versions & slices official artifact (through April 2025) LiveCodeBench · 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 84.7% #1 of 10
GPQA Diamond GPQA Diamond: Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 82.7% #101 of 315
Aider Polyglottwo attempts Aider Polyglot · two attempts: Models edit code for 225 Exercism exercises in six languages, with a second attempt after test feedback Aider 81.3% #3 of 15
No benchmark matches
CheaperR1 0528unstatedOpen-12%-80%$0.39 / $1.63Context: 200K → 128K+1
OpenKimi K2 0711unstated-27%-75%$0.50 / $2Context: 200K → 128K+2
Qwen3 32BunstatedOpen-51%-65%$0.70 / $2.80Context: 200K → 33K+1