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 14.7 #150 of 339
Knowledge 7
AA OmniscienceAccuracy Measures correct answers on the AA Omniscience factual knowledge test AA 20% #157 of 323+1 versions & slices Hallucination Rate Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test AA 85% #202 of 323
AA Omniscience Index Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty AA -47.9 #201 of 323
IFBench (instructions) Checks whether the model follows precise formatting and content instructions AA 75.6% #21 of 248
EQ-Bench3 Judge-rated responses to social and emotionally nuanced scenarios, scored through the source pairwise ranking EQ-Bench 1,464 #23 of 53
SimpleQA Verifiedv1 Accuracy on short factual questions with verified answers, answered without web access 10.4% #79 of 84+1 versions & slices v1 · October 2026 run Accuracy on short factual questions with verified answers, answered without web access 10.7% #1 of 1
Science 3
Humanity's Last Exam Very hard expert questions across many fields, written to stump today’s models AA 23.6% #107 of 339
CritPt (physics) Research-level physics problems that need many careful reasoning steps AA 1.4% #115 of 326
SciCode Writing code for real scientific computing problems (physics, biology, math) AA 45.5% #73 of 142
Coding 8
Terminal-Bench4.0 Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite 0% #77 of 134+1 versions & slices 2.1 Older, easier version of Terminal-Bench; Still widely reported AA 43.4% #81 of 156
DeepSWE1.1 Agents change repository code to resolve software engineering tasks under the stated DeepSWE edition 0% #41 of 44
ITBench SRE IT incident tasks: diagnose and fix broken systems like a site-reliability engineer AA 37.3% #22 of 37
WebDev Arena Human pairwise preferences in the stated webdev vote pool Arena 1,367 #76 of 97+2 versions & slices October 23, 2026 Human pairwise preferences in the stated webdev vote pool Arena 1,368 #77 of 98October 19, 2026 Human pairwise preferences in the stated webdev vote pool Arena 1,367 #76 of 97
Terminal-Bench Hard Agents solve difficult software, system-administration and data-processing tasks in a terminal AA 36.4% #39 of 242
Agents 5
AutomationBench-AA Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation AA 4.9% #87 of 135
AA-Briefcase1.1 · Elo Agents produce professional deliverables; blind pairwise comparisons produce the source Elo rating AA 368 #107 of 142
EnterpriseOps-Gym-AA Agents complete stateful enterprise workflows in eight domains; final database state determines success AA 28.3% #38 of 43
GDPval-AA (work tasks)2.1 Real work tasks from many professions (reports, spreadsheets, analyses), graded against expert work AA 619 #95 of 177
τ³-BankingAA Agents retrieve banking knowledge and execute tools to resolve customer-support workflows AA 14.8% #71 of 134
Multimodal 2
MMMU-Pro (vision) College-level questions that require understanding images: charts, diagrams, photos AA 73.4% #62 of 128
Arena Vision Human pairwise preferences in the stated vision vote pool Arena 1,276 #27 of 88
Writing 17
Arena Textcreative writing · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,416 #50 of 197+13 versions & slices creative writing Human pairwise preferences in the stated text vote pool Arena 1,419 #55 of 195overall Human pairwise preferences in the stated text vote pool Arena 1,452 #50 of 196Chinese Human pairwise preferences in the stated text vote pool Arena 1,475 #61 of 198English Human pairwise preferences in the stated text vote pool Arena 1,451 #47 of 205French Human pairwise preferences in the stated text vote pool Arena 1,436 #76 of 164instruction following Human pairwise preferences in the stated text vote pool Arena 1,434 #51 of 205non-English Human pairwise preferences in the stated text vote pool Arena 1,430 #48 of 205Polish Human pairwise preferences in the stated text vote pool Arena 1,431 #51 of 142Russian Human pairwise preferences in the stated text vote pool Arena 1,461 #27 of 203Spanish Human pairwise preferences in the stated text vote pool Arena 1,442 #55 of 168overall · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,443 #44 of 188creative writing · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,415 #49 of 187overall · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,443 #46 of 198
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 56.5% #58 of 96
EQ-Bench Creative Writingv3 Creative Writing v3 · Elo: Judge-rated creative writing under the v3 prompt set and pairwise ranking EQ-Bench 1,369 #62 of 94
Lech Mazur Short-Story Judges compare stories written to ten required elements; the comparison score can be negative Lech -1.9 #35 of 45
Documents 3
AA-LCR (long context)1.1 Questions that need reading and connecting facts across very long documents (around 100K tokens) AA 69.7% #128 of 317
GDP.pdfAA all-pass Models reason over professional PDF documents; an attempt passes only when every atomic criterion passes 6% #80 of 135
Arena Document Human pairwise preferences in the stated document vote pool Arena 1,425 #27 of 33
Legacy · saturated or no longer reported 12
GPQA Diamond Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 85.7% #71 of 315+1 versions & slices Epoch Models answer graduate-level science multiple-choice questions Epoch 75.8% #74 of 134
Video-MME2 · without subtitles/audio Grouped non-linear question scoring for video understanding; this is distinct from mean question accuracy Video-MME 19% #5 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 28.2% #5 of 18
OTIS Mock AIME 20242025 Models solve the stated OTIS mock AIME mathematics problems Epoch 73.3% #74 of 131
SWE-rebenchall · February 1, 2026 to May 15, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 25.1% #4 of 4+5 versions & slices Go · February 1, 2026 to May 15, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 4Java · February 1, 2026 to May 15, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 4Python · February 1, 2026 to May 15, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 25.1% #4 of 4Rust · February 1, 2026 to May 15, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 4TypeScript · February 1, 2026 to May 15, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 4
τ²-benchAA Tests service agents that use tools while coordinating with a simulated user AA 59.9% #104 of 245
No benchmark matches
CheaperNemotron 3.5 Super VLthinkingOpen+88%-17%$0.19Memory (4-bit): 19 GB → 75 GB
BetterGPT-6 Lunalow+68%+25%$0.005Function calling only with none · Chat Completions+3
OpenQwen3.8 27Bxhigh×2.3+498%$1.15max_tokens range · QwenCloud+4
CheaperLing 3.0 Flash VLthinkingOpen+88%-44%$0.06 / $0.18Memory (4-bit): 19 GB → 75 GB
Solar Mini 4+77%-45%$0.51No image input
Claude Haiku 5.5low+83%+25%$0.024Remove sampling parameters+5
MiMo v2.5thinkingOpen+61%+9%$0.042Memory (4-bit): 19 GB → 186 GB