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 22.9 #98 of 339
Knowledge 4
AA OmniscienceAccuracy Measures correct answers on the AA Omniscience factual knowledge test AA 39.5% #49 of 323+1 versions & slices Hallucination Rate Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test AA 83.2% #191 of 323
AA Omniscience Index Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty AA -10.9 #99 of 323
IFBench (instructions) Checks whether the model follows precise formatting and content instructions AA 70.6% #48 of 248
Reasoning 4
ARC-AGI-1public evaluation Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 65.9% #46 of 64+1 versions & slices semi-private Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 65.7% #44 of 67
ARC-AGI-2public evaluation Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 9.6% #45 of 66+1 versions & slices semi-private Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 9.9% #49 of 69
Math 15
MathArenaAIME 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 95% #6 of 30+14 versions & slices Apex Answer accuracy on the named mathematics competition under the recorded effort MathArena 1% #25 of 34BRUMO 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 91.7% #12 of 24CMIMC 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 88.8% #8 of 22Project Euler Answer accuracy on the named mathematics competition under the recorded effort MathArena 56.8% #9 of 11HMMT February 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 88.3% #10 of 30HMMT November 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 89.2% #7 of 14IMO 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 38.1% #1 of 4Kangaroo 2025 1-2 Answer accuracy on the named mathematics competition under the recorded effort MathArena 68.8% #11 of 17Kangaroo 2025 11-12 Answer accuracy on the named mathematics competition under the recorded effort MathArena 95% #9 of 17Kangaroo 2025 3-4 Answer accuracy on the named mathematics competition under the recorded effort MathArena 60.4% #14 of 17Kangaroo 2025 5-6 Answer accuracy on the named mathematics competition under the recorded effort MathArena 65% #14 of 17Kangaroo 2025 7-8 Answer accuracy on the named mathematics competition under the recorded effort MathArena 90.8% #5 of 16Kangaroo 2025 9-10 Answer accuracy on the named mathematics competition under the recorded effort MathArena 92.5% #12 of 16SMT 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 92% #3 of 24
Science 2
Humanity's Last Exam Very hard expert questions across many fields, written to stump today’s models AA 25.4% #101 of 339
CritPt (physics) Research-level physics problems that need many careful reasoning steps AA 0% #181 of 326
Coding 5
WebDev Arena Human pairwise preferences in the stated webdev vote pool Arena 1,418 #62 of 97+2 versions & slices October 23, 2026 Human pairwise preferences in the stated webdev vote pool Arena 1,418 #63 of 98October 19, 2026 Human pairwise preferences in the stated webdev vote pool Arena 1,418 #62 of 97
SWE-bench Pro (private) Agents resolve long-horizon professional repository tasks; public and private splits are separate Scale 14.9% #8 of 9
Terminal-Bench Hard Agents solve difficult software, system-administration and data-processing tasks in a terminal AA 37.9% #36 of 242
Agents 2
GDPval-AA (work tasks)2.1 Real work tasks from many professions (reports, spreadsheets, analyses), graded against expert work AA 20.9 #136 of 177
τ³-BankingAA Agents retrieve banking knowledge and execute tools to resolve customer-support workflows AA 22.1% #48 of 134
Multimodal 2
MMMU-Pro (vision) College-level questions that require understanding images: charts, diagrams, photos AA 74.3% #56 of 128
Arena Vision Human pairwise preferences in the stated vision vote pool Arena 1,208 #62 of 88
Writing 17
Arena Textcreative writing · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,364 #88 of 197+16 versions & slices creative writing Human pairwise preferences in the stated text vote pool Arena 1,372 #87 of 195overall Human pairwise preferences in the stated text vote pool Arena 1,435 #70 of 196Chinese Human pairwise preferences in the stated text vote pool Arena 1,422 #104 of 198English Human pairwise preferences in the stated text vote pool Arena 1,409 #99 of 205French Human pairwise preferences in the stated text vote pool Arena 1,408 #96 of 164German Human pairwise preferences in the stated text vote pool Arena 1,416 #69 of 167instruction following Human pairwise preferences in the stated text vote pool Arena 1,388 #94 of 205Japanese Human pairwise preferences in the stated text vote pool Arena 1,402 #49 of 152Korean Human pairwise preferences in the stated text vote pool Arena 1,359 #80 of 156non-English Human pairwise preferences in the stated text vote pool Arena 1,397 #89 of 205Polish Human pairwise preferences in the stated text vote pool Arena 1,412 #68 of 142Russian Human pairwise preferences in the stated text vote pool Arena 1,402 #84 of 203Spanish Human pairwise preferences in the stated text vote pool Arena 1,388 #101 of 168overall · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,406 #87 of 188creative writing · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,364 #86 of 187overall · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,406 #89 of 198
Health & law 3
Vals LegalBench Legal reasoning and legal-language understanding on LegalBench Vals 86% #14 of 89
Vals MedCode Medical coding accuracy on patient encounters Vals 49.6% #17 of 78
Vals MedScribe Accuracy of clinical notes produced from consultation transcripts Vals 83.7% #38 of 80
Documents 1
AA-LCR (long context)1.1 Questions that need reading and connecting facts across very long documents (around 100K tokens) AA 76% #85 of 317
Legacy · saturated or no longer reported 13
SWE-bench Pro (public) Agents resolve long-horizon professional repository tasks; public and private splits are separate 41.8% #8 of 19
AIME2025 · AA archived AA evaluates the 30 integer-answer problems from the 2025 AIME mathematics competition AA 91.7% #10 of 140
LiveCodeBenchAA archived AA evaluates solutions to newly collected competitive-programming problems AA 70.3% #44 of 153
GPQA Diamond Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 84.2% #89 of 315
Aider Polyglottwo attempts Models edit code for 225 Exercism exercises in six languages, with a second attempt after test feedback Aider 86.7% #1 of 15
τ²-benchairline · 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 62.5% #3 of 8+4 versions & slices overall · 0.1.3 τ-bench official · 2-overall-0.1.3: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools 80% #3 of 7retail · 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 81.6% #2 of 8telecom · 0.1.3 τ-bench official · 2-telecom-0.1.3: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools 95.8% #3 of 7AA Tests service agents that use tools while coordinating with a simulated user AA 86.5% #57 of 245
Video-MME2 · without subtitles/audio Grouped non-linear question scoring for video understanding; this is distinct from mean question accuracy Video-MME 26.4% #4 of 18+1 versions & slices 2 · with subtitles/audio Grouped non-linear question scoring for video understanding; this is distinct from mean question accuracy Video-MME 37% #3 of 18
Terminal-Bench2.1 Older, easier version of Terminal-Bench; Still widely reported AA 35.2% #92 of 156
No benchmark matches
CheaperGLM 5.1unstatedOpensame-37%$1.40 / $4.40Context: 400K → 203K+1
BetterGemini 3.1 Pro Previewunstatedwins 55%+31%$2 / $12Cannot use both thinking controls
CheaperKimi K2.5unstatedOpensame-65%$0.60 / $3Context: 400K → 256K
Claude Opus 4.6unstatedwins 54%+191%$5 / $25