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 #157 of 339
Knowledge 5
AA OmniscienceAccuracy Measures correct answers on the AA Omniscience factual knowledge test AA 23.7% #126 of 323+1 versions & slices Hallucination Rate Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test AA 74.5% #143 of 323
AA Omniscience Index Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty AA -33.1 #158 of 323
IFBench (instructions) Checks whether the model follows precise formatting and content instructions AA 47.6% #124 of 248
SimpleQA Verifiedv1 Accuracy on short factual questions with verified answers, answered without web access 33.2% #54 of 84
Reasoning 1
Chess Puzzles Models choose the best next move from a novel chess position Epoch 18% #57 of 69
Math 3
AA Analyst Agent Data-analyst tasks: the model explores data with tools and reports findings AA 8.8% #40 of 43
FrontierMath Tier 4 (v2) Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 14.6% #47 of 55
FrontierMath Tiers 1–3 (v2) Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 42.8% #51 of 70
Science 3
Humanity's Last Exam Very hard expert questions across many fields, written to stump today’s models AA 6.8% #232 of 339
CritPt (physics) Research-level physics problems that need many careful reasoning steps AA 0% #181 of 326
SciCode Writing code for real scientific computing problems (physics, biology, math) AA 39.4% #100 of 142
Coding 7
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 34.1% #93 of 156
ITBench SRE IT incident tasks: diagnose and fix broken systems like a site-reliability engineer AA 32.7% #27 of 37
WebDev Arena Human pairwise preferences in the stated webdev vote pool Arena 1,358 #79 of 97+2 versions & slices October 23, 2026 Human pairwise preferences in the stated webdev vote pool Arena 1,357 #81 of 98October 19, 2026 Human pairwise preferences in the stated webdev vote pool Arena 1,358 #80 of 97
Terminal-Bench Hard Agents solve difficult software, system-administration and data-processing tasks in a terminal AA 18.9% #94 of 242
Agents 6
AutomationBench-AA Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation AA 0.9% #114 of 135
AA-Briefcase1.1 · Elo Agents produce professional deliverables; blind pairwise comparisons produce the source Elo rating AA 724 #83 of 142
APEX-Agents Agents complete long-horizon professional-services tasks across realistic applications AA 17% #19 of 29
EnterpriseOps-Gym-AA Agents complete stateful enterprise workflows in eight domains; final database state determines success AA 26.3% #40 of 43
GDPval-AA (work tasks)2.1 Real work tasks from many professions (reports, spreadsheets, analyses), graded against expert work AA 941 #79 of 177
τ³-BankingAA Agents retrieve banking knowledge and execute tools to resolve customer-support workflows AA 8% #99 of 134
Multimodal 3
MMMU-Pro (vision) College-level questions that require understanding images: charts, diagrams, photos AA 64.8% #83 of 128
Arena Vision Human pairwise preferences in the stated vision vote pool Arena 1,229 #55 of 88
Chartography Expert chart reasoning over professional scientific, medical, financial and engineering charts Surge 12.9% #32 of 37
Writing 18
Arena Textcreative writing · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,380 #80 of 197+16 versions & slices creative writing Human pairwise preferences in the stated text vote pool Arena 1,422 #52 of 195overall Human pairwise preferences in the stated text vote pool Arena 1,442 #61 of 196Chinese Human pairwise preferences in the stated text vote pool Arena 1,421 #105 of 198English Human pairwise preferences in the stated text vote pool Arena 1,404 #102 of 205French Human pairwise preferences in the stated text vote pool Arena 1,412 #93 of 164German Human pairwise preferences in the stated text vote pool Arena 1,395 #79 of 167instruction following Human pairwise preferences in the stated text vote pool Arena 1,365 #108 of 205Japanese Human pairwise preferences in the stated text vote pool Arena 1,379 #64 of 152Korean Human pairwise preferences in the stated text vote pool Arena 1,355 #84 of 156non-English Human pairwise preferences in the stated text vote pool Arena 1,385 #95 of 205Polish Human pairwise preferences in the stated text vote pool Arena 1,397 #77 of 142Russian Human pairwise preferences in the stated text vote pool Arena 1,399 #87 of 203Spanish Human pairwise preferences in the stated text vote pool Arena 1,399 #93 of 168overall · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,397 #93 of 188creative writing · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,380 #78 of 187overall · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,397 #98 of 198
Lech Mazur Short-Story Judges compare stories written to ten required elements; the comparison score can be negative Lech -4.3 #44 of 45
Health & law 5
MLCR-AA Models synthesize long medical records; the headline credits concise answers judged both accurate and complete AA 5.6% #75 of 103
Vals LegalBench Legal reasoning and legal-language understanding on LegalBench Vals 84.5% #28 of 89
Vals MedCode Medical coding accuracy on patient encounters Vals 38.1% #56 of 78
Vals MedScribe Accuracy of clinical notes produced from consultation transcripts Vals 74.4% #62 of 80
Vals Legal Research Bench Grounded legal research; a task passes only if all criteria pass Vals 15.4% #58 of 73
Documents 2
AA-LCR (long context)1.1 Questions that need reading and connecting facts across very long documents (around 100K tokens) AA 32.3% #218 of 317
GDP.pdfAA all-pass Models reason over professional PDF documents; an attempt passes only when every atomic criterion passes 2.8% #103 of 135
Legacy · saturated or no longer reported 4
GPQA Diamond Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 65.8% #210 of 315+1 versions & slices Epoch Models answer graduate-level science multiple-choice questions Epoch 88.8% #39 of 134
OTIS Mock AIME 20242025 Models solve the stated OTIS mock AIME mathematics problems Epoch 93.3% #39 of 131
τ²-benchAA Tests service agents that use tools while coordinating with a simulated user AA 65.8% #97 of 245
No benchmark matches
CheaperGPT-6.1 Sollow+70.7-11%$0.13Remove unsupported parameters+5
OpenMuse Spark Openxhigh+57.7+477%$0.84Reasoning off leaves output unchanged · Self-hosted / llama.cpp+1
Gemini 3 Flash Previewthinking+43.2-28%$0.50 / $3
Granite 4.2 3BthinkingOpen+18.4-96%$0.006Context: 1M → 128K+1
Mistral Small 4thinkingOpen+2.7-88%$0.017Context: 1M → 262K
Kimi K3maxOpen+52.8+1,268%$2.00Sampling parameters are fixed+3
Step 5 PreviewthinkingOpen+49.5+602%$1.03web_search not available · Chat Completions+2
GLM 5.3maxOpen+47.4+1,272%$2.01No image input+4MMotif 3Open+43.3-44%$0.50 / $2Context: 1M → 262K+1