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 30.9 #53 of 337
Knowledge 7
AA OmniscienceAccuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test AA 44.7% #34 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 54.1% #97 of 321
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty AA 14.8 #31 of 321
IFBench (instructions) IFBench (instructions): Checks whether the model follows precise formatting and content instructions AA 43.6% #137 of 248
EQ-Bench3 EQ-Bench 3: Judge-rated responses to social and emotionally nuanced scenarios, scored through the source pairwise ranking EQ-Bench 1,884 #3 of 50
SimpleQA Verifiedv1 SimpleQA Verified v1: Accuracy on short factual questions with verified answers, answered without web access 51.7% #26 of 83
Mystery Game Puzzles Mystery Game Puzzles: Models choose the best move in puzzle positions from an undisclosed game Epoch 28% #27 of 64
Reasoning 1
Chess Puzzles Chess Puzzles: Models choose the best next move from a novel chess position Epoch 41% #28 of 68
Math 17
AA Analyst Agent AA Analyst Agent: Data-analyst tasks: the model explores data with tools and reports findings AA 43.8% #17 of 43
MathArenaAIME 2026 MathArena · AIME 2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 95.8% #12 of 24+11 versions & slices HMMT February 2026 MathArena · HMMT February 2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 93.9% #9 of 24Apex MathArena · apex · Apex: Answer accuracy on the named mathematics competition under the recorded effort MathArena 40.6% #7 of 34Apex Shortlist MathArena · apex · Apex Shortlist: Answer accuracy on the named mathematics competition under the recorded effort MathArena 63.8% #18 of 30ArXivMath · April 2026 MathArena · arxiv · 04/2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 57.5% #9 of 17ArXivMath · February 2026 MathArena · arxiv · 02/2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 40.6% #11 of 20ArXivMath · January 2026 MathArena · arxiv · 01/2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 52.2% #14 of 18ArXivMath · March 2026 MathArena · arxiv · 03/2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 50.8% #10 of 14BrokenArXiv · April 2026 MathArena · arxiv_false · 04/2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 4.1% #15 of 15BrokenArXiv · February 2026 MathArena · arxiv_false · 02/2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 4% #13 of 15BrokenArXiv · March 2026 MathArena · arxiv_false · 03/2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 5.8% #12 of 13ArXivLean · March 2026 MathArena · arxivlean · 03/2026: Answer accuracy on the named mathematics competition under the recorded effort MathArena 14.6% #3 of 5
FrontierMath FrontierMath: Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 43.8% #4 of 40
FrontierMath Tier4 FrontierMath Tier 4: Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 22.9% #3 of 33
FrontierMath Tier 4 (v2) FrontierMath Tier 4 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 31.7% #25 of 54
FrontierMath Tiers 1–3 (v2) FrontierMath Tiers 1–3 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 70.2% #23 of 69
Science 2
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models AA 33.3% #71 of 337
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps AA 5.1% #78 of 324
Coding 7
DeepSWE1.1 DeepSWE 1.1: Agents change repository code to resolve software engineering tasks under the stated DeepSWE edition 49.3% #27 of 44+1 versions & slices 1 DeepSWE 1: Agents change repository code to resolve software engineering tasks under the stated DeepSWE edition Datacurve 40.3% #5 of 20
ITBench SRE ITBench SRE: IT incident tasks: diagnose and fix broken systems like a site-reliability engineer AA 46.7% #10 of 37
WebDev Arena WebDev Arena: Human pairwise preferences in the stated webdev vote pool Arena 1,558 #32 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 1,558 #32 of 97
MirrorCode MirrorCode: Agents reimplement software projects by observing their behavior Epoch 31.1% #5 of 9
Terminal-Bench Hard Terminal-Bench Hard (version unstated): Agents solve difficult software, system-administration and data-processing tasks in a terminal AA 54.5% #7 of 242
Agents 8
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 42.6 #43 of 175
AA-Briefcase1.1 · Elo AA-Briefcase 1.1 · Elo: Agents produce professional deliverables; blind pairwise comparisons produce the source Elo rating AA 1,068 #50 of 140
MCP-Atlas Public MCPAtlas Public: MCPAtlas Public as reported in this launch evaluation; the post is the authority for the recorded task and grading scope 80.8% #2 of 16
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 40.2% #6 of 21
EBR Bench EBR Bench: Agents learn to play Earthborne Rangers across repeated playthroughs Epoch 19% #17 of 24
Furniture Assembly Furniture Assembly: Models inspect furniture assembly photos, identify incorrect builds and locate the mistaken step Epoch 33.3% #22 of 31
MCP-Atlasfull 1000 MCP-Atlas · full 1000: Tool orchestration across MCP servers on the stated task split Scale 79.1% #2 of 16
τ³-BankingAA τ³-Banking · AA: Agents retrieve banking knowledge and execute tools to resolve customer-support workflows AA 34.6% #26 of 134
Multimodal 3
MMMU-Pro (vision) MMMU-Pro (vision): College-level questions that require understanding images: charts, diagrams, photos AA 76.4% #39 of 127
Arena Vision Arena Vision · overall: Human pairwise preferences in the stated vision vote pool Arena 1,313 #7 of 88
Chartography Chartography: Expert chart reasoning over professional scientific, medical, financial and engineering charts Surge 16.5% #26 of 37
Writing 18
Arena Textcreative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool Arena 1,489 #5 of 195+14 versions & slices overall Arena Text · overall: Human pairwise preferences in the stated text vote pool Arena 1,501 #7 of 196Chinese Arena Text · chinese: Human pairwise preferences in the stated text vote pool Arena 1,532 #11 of 181English Arena Text · english: Human pairwise preferences in the stated text vote pool Arena 1,496 #10 of 188French Arena Text · french: Human pairwise preferences in the stated text vote pool Arena 1,503 #7 of 149German Arena Text · german: Human pairwise preferences in the stated text vote pool Arena 1,495 #8 of 154instruction following Arena Text · instruction following: Human pairwise preferences in the stated text vote pool Arena 1,498 #7 of 188Japanese Arena Text · japanese: Human pairwise preferences in the stated text vote pool Arena 1,474 #13 of 138Korean Arena Text · korean: Human pairwise preferences in the stated text vote pool Arena 1,451 #14 of 144non-English Arena Text · non english: Human pairwise preferences in the stated text vote pool Arena 1,480 #10 of 188Polish Arena Text · polish: Human pairwise preferences in the stated text vote pool Arena 1,480 #10 of 133Russian Arena Text · russian: Human pairwise preferences in the stated text vote pool Arena 1,494 #10 of 186Spanish Arena Text · spanish: Human pairwise preferences in the stated text vote pool Arena 1,495 #8 of 155overall · full cohort Arena Text · overall · snapshot 2026-10-15 · native board settings: Human pairwise preferences in the stated text vote pool Arena 1,490 #11 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 1,486 #9 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 81.8% #10 of 95
EQ-Bench Creative Writingv3 Creative Writing v3 · Elo: Judge-rated creative writing under the v3 prompt set and pairwise ranking EQ-Bench 1,914 #13 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 1.8 #12 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 28.9% #17 of 64
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench Vals 85.3% #20 of 89
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters Vals 54.9% #6 of 78
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts Vals 83% #42 of 80
Vals Legal Research Bench Vals Legal Research Bench: Grounded legal research; a task passes only if all criteria pass Vals 38.5% #28 of 69
Documents 2
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 75.7% #85 of 275
Arena Document Arena Document · overall: Human pairwise preferences in the stated document vote pool Arena 1,495 #5 of 33
Legacy · saturated or no longer reported 12
GPQA Diamond GPQA Diamond: Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 88.5% #48 of 315+1 versions & slices Epoch GPQA Diamond · source setting 2: Models answer graduate-level science multiple-choice questions Epoch 90.2% #32 of 133
OTIS Mock AIME 20242025 OTIS Mock AIME 2024 2025: Models solve the stated OTIS mock AIME mathematics problems Epoch 97.8% #25 of 130
SWE-bench VerifiedEpoch Swe Bench Verified: Agents modify real repositories to resolve the stated Verified issues Epoch 83.5% #1 of 24
SWE-rebenchall · 2026-03-01 to 2026-05-15 SWE-rebench · all · 2026-03-01 to 2026-05-15: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 53.1% #2 of 5+5 versions & slices Go · 2026-03-01 to 2026-05-15 SWE-rebench · go · 2026-03-01 to 2026-05-15: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 5Java · 2026-03-01 to 2026-05-15 SWE-rebench · java · 2026-03-01 to 2026-05-15: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 5Python · 2026-03-01 to 2026-05-15 SWE-rebench · python · 2026-03-01 to 2026-05-15: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 53.1% #2 of 5Rust · 2026-03-01 to 2026-05-15 SWE-rebench · rust · 2026-03-01 to 2026-05-15: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 5TypeScript · 2026-03-01 to 2026-05-15 SWE-rebench · typescript · 2026-03-01 to 2026-05-15: Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 5
τ²-benchAA τ²-Bench: Tests service agents that use tools while coordinating with a simulated user AA 74% #91 of 245
Terminal-Bench2.1 Terminal-Bench 2.1 · source setting 2: Older, easier version of Terminal-Bench; Still widely reported AA 83.1% #17 of 156
No benchmark matches
BetterGPT-6 Lunaxhigh+12%-98%$0.042Function calling only with none · Chat Completions+3
OpenMiMo v2.6 Prothinking+50%-95%$0.13Pass back reasoning_content+3
CheaperGemini 4 Argonhigh+70%-60%*$1.99Synchronous output stops at 65,536+2
Muse Wondermax+69%-20%$4.52Only auto tool_choice is supported+5
DeepSeek V4.1 Flash 1017maxOpen+46%-95%$0.26Pass back reasoning_content · Chat Completions+4
GLM 5.3maxOpen+45%-79%$2.01No image input+4
Step 5 PreviewthinkingOpen+41%-86%$1.03web_search not available · Chat Completions+2
Kimi K3maxOpen+41%-40%$2.00Sampling parameters are fixed+3