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 16.1 #139 of 337
Knowledge 4
AA OmniscienceAccuracy Measures correct answers on the AA Omniscience factual knowledge test AA 39.1% #50 of 321+1 versions & slices Hallucination Rate Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test AA 90.9% #258 of 321
AA Omniscience Index Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty AA -16.4 #110 of 321
IFBench (instructions) Checks whether the model follows precise formatting and content instructions AA 48.7% #122 of 248
Reasoning 4
ARC-AGI-1public evaluation Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 56.4% #50 of 61+1 versions & slices semi-private Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 41% #53 of 64
ARC-AGI-2public evaluation Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 5.1% #47 of 63+1 versions & slices semi-private Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 4.9% #49 of 66
Math 20
FrontierMath Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 14.1% #24 of 40
FrontierMath Tier4 Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 4.2% #16 of 33
FrontierMath Tier 4 (v2) Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 0% #53 of 54
FrontierMath Tiers 1–3 (v2) Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 24.6% #60 of 69
MathArenaAIME 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 88.3% #15 of 30+15 versions & slices Apex Answer accuracy on the named mathematics competition under the recorded effort MathArena 0.5% #31 of 34BRUMO 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 90% #14 of 24CMIMC 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 58.1% #21 of 22Project Euler Answer accuracy on the named mathematics competition under the recorded effort MathArena 26.7% #11 of 11HMMT February 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 82.5% #13 of 30HMMT November 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 80% #12 of 14IMO 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 31.6% #2 of 4Kangaroo 2025 1-2 Answer accuracy on the named mathematics competition under the recorded effort MathArena 64.6% #14 of 17Kangaroo 2025 11-12 Answer accuracy on the named mathematics competition under the recorded effort MathArena 89.2% #13 of 17Kangaroo 2025 3-4 Answer accuracy on the named mathematics competition under the recorded effort MathArena 64.6% #12 of 17Kangaroo 2025 5-6 Answer accuracy on the named mathematics competition under the recorded effort MathArena 66.7% #13 of 17Kangaroo 2025 7-8 Answer accuracy on the named mathematics competition under the recorded effort MathArena 82.5% #14 of 16Kangaroo 2025 9-10 Answer accuracy on the named mathematics competition under the recorded effort MathArena 95.8% #9 of 16SMT 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 84.9% #14 of 24USAMO 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 24.4% #2 of 3
Science 3
Humanity's Last Exam Very hard expert questions across many fields, written to stump today’s models AA 22.5% #111 of 337
CritPt (physics) Research-level physics problems that need many careful reasoning steps AA 2.6% #97 of 324
SciCode Writing code for real scientific computing problems (physics, biology, math) AA 46.3% #68 of 140
Coding 5
Terminal-Bench4.0 Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite 0% #75 of 132+1 versions & slices 2.1 Older, easier version of Terminal-Bench; Still widely reported AA 28.5% #99 of 156
WebDev Arena Human pairwise preferences in the stated webdev vote pool Arena 1,226 #94 of 97+1 versions & slices full cohort Human pairwise preferences in the stated webdev vote pool Arena 1,226 #94 of 97
Terminal-Bench Hard Agents solve difficult software, system-administration and data-processing tasks in a terminal AA 26.5% #68 of 242
Agents 5
AutomationBench-AA Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation AA 2.2% #96 of 133
AA-Briefcase1.1 · Elo Agents produce professional deliverables; blind pairwise comparisons produce the source Elo rating AA 302 #110 of 140
GDPval-AA (work tasks)2.1 Real work tasks from many professions (reports, spreadsheets, analyses), graded against expert work AA 0 #122 of 175
τ³-benchbanking · 1.0.1 τ-bench official · 3-banking-1.0.1: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools Sierra 13.7% #19 of 21
τ³-BankingAA Agents retrieve banking knowledge and execute tools to resolve customer-support workflows AA 9.7% #92 of 134
Multimodal 2
MMMU-Pro (vision) College-level questions that require understanding images: charts, diagrams, photos AA 74.9% #52 of 127
Arena Vision Human pairwise preferences in the stated vision vote pool Arena 1,263 #36 of 88
Writing 15
Arena Textcreative writing Human pairwise preferences in the stated text vote pool Arena 1,443 #34 of 195+14 versions & slices overall Human pairwise preferences in the stated text vote pool Arena 1,445 #57 of 196Chinese Human pairwise preferences in the stated text vote pool Arena 1,507 #34 of 181English Human pairwise preferences in the stated text vote pool Arena 1,458 #36 of 188French Human pairwise preferences in the stated text vote pool Arena 1,472 #28 of 149German Human pairwise preferences in the stated text vote pool Arena 1,487 #12 of 154instruction following Human pairwise preferences in the stated text vote pool Arena 1,436 #49 of 188Japanese Human pairwise preferences in the stated text vote pool Arena 1,461 #15 of 138Korean Human pairwise preferences in the stated text vote pool Arena 1,435 #21 of 144non-English Human pairwise preferences in the stated text vote pool Arena 1,451 #27 of 188Polish Human pairwise preferences in the stated text vote pool Arena 1,479 #11 of 133Russian Human pairwise preferences in the stated text vote pool Arena 1,460 #28 of 186Spanish Human pairwise preferences in the stated text vote pool Arena 1,473 #15 of 155overall · full cohort Human pairwise preferences in the stated text vote pool Arena 1,458 #31 of 188creative writing · full cohort Human pairwise preferences in the stated text vote pool Arena 1,454 #22 of 187
Health & law 2
Vals MedCode Medical coding accuracy on patient encounters Vals 50.6% #14 of 78
Vals MedScribe Accuracy of clinical notes produced from consultation transcripts Vals 73.6% #63 of 80
Documents 3
AA-LCR (long context)1.1 Questions that need reading and connecting facts across very long documents (around 100K tokens) AA 69% #131 of 275
GDP.pdfAA all-pass Models reason over professional PDF documents; an attempt passes only when every atomic criterion passes 10.2% #59 of 133
Arena Document Human pairwise preferences in the stated document vote pool Arena 1,421 #28 of 33
Other 1
AA Speech to textRecorded audio word error rate Measures transcription errors as a share of reference words in the non-streaming test AA 2.9% #13 of 55
Legacy · saturated or no longer reported 13
AIME2025 · AA archived AA evaluates the 30 integer-answer problems from the 2025 AIME mathematics competition AA 87.7% #26 of 140
LiveCodeBenchAA archived AA evaluates solutions to newly collected competitive-programming problems AA 80.1% #16 of 153
GPQA Diamond Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 84.4% #88 of 315+1 versions & slices Epoch Models answer graduate-level science multiple-choice questions Epoch 85.3% #56 of 133
OTIS Mock AIME 20242025 Models solve the stated OTIS mock AIME mathematics problems Epoch 84.2% #66 of 130
SWE-bench VerifiedEpoch Swe Bench Verified: Agents modify real repositories to resolve the stated Verified issues Epoch 57.6% #22 of 24
SWE-rebenchall · 2025-07-01 to 2025-11-01 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 23.5% #1 of 2+5 versions & slices Go · 2025-07-01 to 2025-11-01 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 2Java · 2025-07-01 to 2025-11-01 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 2Python · 2025-07-01 to 2025-11-01 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 23.5% #1 of 2Rust · 2025-07-01 to 2025-11-01 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 2TypeScript · 2025-07-01 to 2025-11-01 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 2
τ²-benchAA Tests service agents that use tools while coordinating with a simulated user AA 54.1% #107 of 245
No benchmark matches
CheaperGPT-5.6 Lunamax×3.3-87%$0.18
OpenKimi K3max×2.9+75%$2.00Sampling parameters are fixed+3
CheaperClaude Haiku 5.5max×3.1-94%$0.21Remove sampling parameters+5
CheaperMuse Spark 1.3xhigh×3-42%$1.37Only auto tool_choice is supported+5
Grok 4.3high+74%-55%$0.21
GLM 5.3maxOpen×2.8-37%$2.01No image input+4
DeepSeek V4 Pro 0813maxOpen×2.6-42%$0.67No image input+5