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 26 #77 of 337
Knowledge 4
AA OmniscienceAccuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test AA 22.4% #134 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 24.7% #18 of 321
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty AA 3.3 #58 of 321
IFBench (instructions) IFBench (instructions): Checks whether the model follows precise formatting and content instructions AA 79.9% #6 of 248
Math 1
AA Analyst Agent AA Analyst Agent: Data-analyst tasks: the model explores data with tools and reports findings AA 20% #30 of 43
Science 3
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models AA 35.7% #59 of 337
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps AA 4% #87 of 324
SciCode SciCode: Writing code for real scientific computing problems (physics, biology, math) AA 50.6% #48 of 140
Coding 7
Terminal-Bench4.0 Terminal-Bench 4.0 · source setting 2: 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 Terminal-Bench 2.1 · source setting 2: Older, easier version of Terminal-Bench; Still widely reported AA 65.2% #43 of 156
ITBench SRE ITBench SRE: IT incident tasks: diagnose and fix broken systems like a site-reliability engineer AA 38.2% #20 of 37
WebDev Arena WebDev Arena: Human pairwise preferences in the stated webdev vote pool Arena 1,480 #50 of 97+1 versions & slices full cohort WebDev Arena · snapshot 2026-10-19 · native board settings: Human pairwise preferences in the stated webdev vote pool Arena 1,480 #50 of 97
DeepSWE1 DeepSWE 1: Agents change repository code to resolve software engineering tasks under the stated DeepSWE edition Datacurve 19.5% #11 of 20
Terminal-Bench Hard Terminal-Bench Hard (version unstated): Agents solve difficult software, system-administration and data-processing tasks in a terminal AA 43.2% #25 of 242
Agents 5
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation AA 13.7% #71 of 133
AA-Briefcase1.1 · Elo AA-Briefcase 1.1 · Elo: Agents produce professional deliverables; blind pairwise comparisons produce the source Elo rating AA 883 #67 of 140
APEX-Agents APEX-Agents: Agents complete long-horizon professional-services tasks across realistic applications AA 2.4% #27 of 29
GDPval-AA (work tasks)2.1 GDPval-AA (work tasks) 2.1: Real work tasks from many professions (reports, spreadsheets, analyses), graded against expert work AA 31 #60 of 175
τ³-BankingAA τ³-Banking · AA: Agents retrieve banking knowledge and execute tools to resolve customer-support workflows AA 9.9% #90 of 134
Writing 18
Arena Textcreative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool Arena 1,434 #42 of 195+14 versions & slices overall Arena Text · overall: Human pairwise preferences in the stated text vote pool Arena 1,467 #32 of 196Chinese Arena Text · chinese: Human pairwise preferences in the stated text vote pool Arena 1,507 #36 of 181English Arena Text · english: Human pairwise preferences in the stated text vote pool Arena 1,477 #22 of 188French Arena Text · french: Human pairwise preferences in the stated text vote pool Arena 1,488 #18 of 149German Arena Text · german: Human pairwise preferences in the stated text vote pool Arena 1,459 #28 of 154instruction following Arena Text · instruction following: Human pairwise preferences in the stated text vote pool Arena 1,476 #19 of 188Japanese Arena Text · japanese: Human pairwise preferences in the stated text vote pool Arena 1,410 #40 of 138Korean Arena Text · korean: Human pairwise preferences in the stated text vote pool Arena 1,435 #20 of 144non-English Arena Text · non english: Human pairwise preferences in the stated text vote pool Arena 1,449 #30 of 188Polish Arena Text · polish: Human pairwise preferences in the stated text vote pool Arena 1,464 #18 of 133Russian Arena Text · russian: Human pairwise preferences in the stated text vote pool Arena 1,450 #35 of 186Spanish Arena Text · spanish: Human pairwise preferences in the stated text vote pool Arena 1,470 #18 of 155overall · full cohort Arena Text · overall · snapshot 2026-10-15 · native board settings: Human pairwise preferences in the stated text vote pool Arena 1,465 #27 of 188creative writing · full cohort Arena Text · creative writing · snapshot 2026-10-15 · native board settings: Human pairwise preferences in the stated text vote pool Arena 1,438 #35 of 187
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 73.7% #29 of 95
EQ-Bench Creative Writingv3 Creative Writing v3 · Elo: Judge-rated creative writing under the v3 prompt set and pairwise ranking EQ-Bench 1,493 #49 of 91
Lech Mazur Short-Story Lech Mazur Short-Story · comparison score: Judges compare stories written to ten required elements; the comparison score can be negative Lech -0.7 #27 of 44
Health & law 5
MLCR-AA MLCR-AA: Models synthesize long medical records; the headline credits concise answers judged both accurate and complete AA 9.4% #42 of 64
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench Vals 77.1% #73 of 89
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters Vals 32.5% #67 of 78
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts Vals 83.7% #36 of 80
Vals Legal Research Bench Vals Legal Research Bench: Grounded legal research; a task passes only if all criteria pass Vals 15.9% #51 of 69
Documents 2
GDP.pdfAA all-pass GDP.pdf · AA all-pass: Models reason over professional PDF documents; an attempt passes only when every atomic criterion passes 4% #91 of 133
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 79.7% #48 of 275
Legacy · saturated or no longer reported 8
GPQA Diamond GPQA Diamond: Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 86.6% #64 of 315
SWE-rebenchall · 2026-03-01 to 2026-07-01 SWE-rebench · all · 2026-03-01 to 2026-07-01: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 44.4% #3 of 6+5 versions & slices Go · 2026-03-01 to 2026-07-01 SWE-rebench · go · 2026-03-01 to 2026-07-01: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 56.2% #2 of 6Java · 2026-03-01 to 2026-07-01 SWE-rebench · java · 2026-03-01 to 2026-07-01: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 42.9% #3 of 6Python · 2026-03-01 to 2026-07-01 SWE-rebench · python · 2026-03-01 to 2026-07-01: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 42.6% #3 of 6Rust · 2026-03-01 to 2026-07-01 SWE-rebench · rust · 2026-03-01 to 2026-07-01: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 44% #3 of 6TypeScript · 2026-03-01 to 2026-07-01 SWE-rebench · typescript · 2026-03-01 to 2026-07-01: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 45.8% #3 of 6
τ²-benchAA τ²-Bench: Tests service agents that use tools while coordinating with a simulated user AA 94.2% #23 of 245
No benchmark matches
CheaperGLM 4.7 FlashthinkingOpen+5%-73%$0.061 / $0.40Context: 1M → 200K
BetterGrok 4.3high+4%+280%$0.21
CheaperStep 3.7 FlashhighOpen+5%-20%$0.20 / $1.15Context: 1M → 256K
CheaperHy3 previewthinkingOpen-2%-48%$0.18 / $0.60Context: 1M → 256K+1
Qwen3.5 4BthinkingOpen-2%-89%$0.03 / $0.15Context: 1M → 262K
Claude Fable 5max+5%+16,074%$8.75CJT 35B Flash+5%——Context: 1M → 256K