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 27.9 #67 of 339
Knowledge 5
AA OmniscienceAccuracy Measures correct answers on the AA Omniscience factual knowledge test AA 26.3% #103 of 323+1 versions & slices Hallucination Rate Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test AA 35.3% #50 of 323
AA Omniscience Index Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty AA 0.3 #70 of 323
IFBench (instructions) Checks whether the model follows precise formatting and content instructions AA 72.3% #38 of 248
EQ-Bench3 Judge-rated responses to social and emotionally nuanced scenarios, scored through the source pairwise ranking EQ-Bench 1,495 #22 of 53
Reasoning 4
ARC-AGI-1public evaluation Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 58.6% #52 of 64+1 versions & slices semi-private Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 44.7% #55 of 67
ARC-AGI-2public evaluation Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 5.4% #48 of 66+1 versions & slices semi-private Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit ARC 4.9% #52 of 69
Math 19
MathArenaAIME 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 96.7% #6 of 24+16 versions & slices HMMT February 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 86.4% #19 of 24AIME 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 96.7% #3 of 30Apex Answer accuracy on the named mathematics competition under the recorded effort MathArena 10.9% #18 of 34Apex Shortlist Answer accuracy on the named mathematics competition under the recorded effort MathArena 68.6% #15 of 30ArXivMath · December 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 38.2% #10 of 13ArXivMath · February 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 41.4% #10 of 20ArXivMath · January 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 54.4% #12 of 18ArXivMath · March 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 45.8% #12 of 14BrokenArXiv · February 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 11.7% #9 of 15BrokenArXiv · March 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 12.1% #8 of 13BRUMO 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 99.2% #2 of 24CMIMC 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 93.1% #4 of 22HMMT February 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 97.5% #2 of 30HMMT November 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 94.2% #2 of 14SMT 2025 Answer accuracy on the named mathematics competition under the recorded effort MathArena 91% #4 of 24USAMO 2026 Answer accuracy on the named mathematics competition under the recorded effort MathArena 35.1% #8 of 8
FrontierMath Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 16.4% #23 of 40
FrontierMath Tier4 Models solve expert-written advanced mathematics problems on the named tier and edition Epoch 2.1% #22 of 33
Science 2
Humanity's Last Exam Very hard expert questions across many fields, written to stump today’s models AA 29.3% #86 of 339
CritPt (physics) Research-level physics problems that need many careful reasoning steps AA 2% #107 of 326
Coding 5
DeepSWE1.1 Agents change repository code to resolve software engineering tasks under the stated DeepSWE edition 2.1% #39 of 44
WebDev Arena Human pairwise preferences in the stated webdev vote pool Arena 1,435 #58 of 97+2 versions & slices October 23, 2026 Human pairwise preferences in the stated webdev vote pool Arena 1,435 #60 of 98October 19, 2026 Human pairwise preferences in the stated webdev vote pool Arena 1,435 #58 of 97
Terminal-Bench Hard Agents solve difficult software, system-administration and data-processing tasks in a terminal AA 43.2% #25 of 242
Agents 2
APEX-Agents Agents complete long-horizon professional-services tasks across realistic applications AA 14.5% #22 of 29
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 9.8% #20 of 21
Writing 19
Arena Textcreative writing · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,438 #35 of 197+16 versions & slices creative writing Human pairwise preferences in the stated text vote pool Arena 1,448 #28 of 195overall Human pairwise preferences in the stated text vote pool Arena 1,457 #42 of 196Chinese Human pairwise preferences in the stated text vote pool Arena 1,511 #27 of 198English Human pairwise preferences in the stated text vote pool Arena 1,458 #38 of 205French Human pairwise preferences in the stated text vote pool Arena 1,455 #54 of 164German Human pairwise preferences in the stated text vote pool Arena 1,447 #40 of 167instruction following Human pairwise preferences in the stated text vote pool Arena 1,428 #57 of 205Japanese Human pairwise preferences in the stated text vote pool Arena 1,417 #39 of 152Korean Human pairwise preferences in the stated text vote pool Arena 1,422 #32 of 156non-English Human pairwise preferences in the stated text vote pool Arena 1,430 #49 of 205Polish Human pairwise preferences in the stated text vote pool Arena 1,434 #47 of 142Russian Human pairwise preferences in the stated text vote pool Arena 1,436 #47 of 203Spanish Human pairwise preferences in the stated text vote pool Arena 1,454 #39 of 168overall · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,446 #42 of 188creative writing · October 15, 2026 Human pairwise preferences in the stated text vote pool Arena 1,438 #36 of 187overall · October 22, 2026 Human pairwise preferences in the stated text vote pool Arena 1,446 #41 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 70.9% #38 of 96
EQ-Bench Creative Writingv3 Creative Writing v3 · Elo: Judge-rated creative writing under the v3 prompt set and pairwise ranking EQ-Bench 1,601 #44 of 94
Health & law 1
Vals LegalBench Legal reasoning and legal-language understanding on LegalBench Vals 84.1% #37 of 89
Documents 1
AA-LCR (long context)1.1 Questions that need reading and connecting facts across very long documents (around 100K tokens) AA 75.7% #88 of 317
Legacy · saturated or no longer reported 15
τ²-benchAA Tests service agents that use tools while coordinating with a simulated user AA 98.2% #8 of 245+4 versions & slices airline · 1.0.1 τ-bench official · 2-airline-1.0.1: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools Sierra 82.5% #3 of 6overall · 1.0.1 τ-bench official · 2-overall-1.0.1: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools Sierra 81% #5 of 6retail · 1.0.1 τ-bench official · 2-retail-1.0.1: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools Sierra 73.7% #5 of 6telecom · 1.0.1 τ-bench official · 2-telecom-1.0.1: Customer-service agents converse with simulated users, retrieve domain knowledge and change state through tools Sierra 86.8% #5 of 6
GPQA Diamond Graduate-level biology, chemistry and physics questions that can’t be solved by a quick web search 82% #107 of 315+1 versions & slices Epoch Models answer graduate-level science multiple-choice questions Epoch 87.8% #48 of 134
OTIS Mock AIME 20242025 Models solve the stated OTIS mock AIME mathematics problems Epoch 80% #72 of 131
SWE-bench VerifiedEpoch Swe Bench Verified: Agents modify real repositories to resolve the stated Verified issues Epoch 72.1% #17 of 24
SWE-rebenchall · January 1, 2026 to March 1, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 53.3% #1 of 5+5 versions & slices Go · January 1, 2026 to March 1, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 5Java · January 1, 2026 to March 1, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 5Python · January 1, 2026 to March 1, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 53.3% #1 of 5Rust · January 1, 2026 to March 1, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 5TypeScript · January 1, 2026 to March 1, 2026 Repository issue resolution over the exact fresh pull-request window, under the source agent version Nebius 0% #1 of 5
No benchmark matches
CheaperDeepSeek V4.1 Flash 1017maxOpen+24%-66%$0.26Memory (4-bit): 452 GB → 458 GB+5
BetterGPT-6 Lunamax+24%-87%$0.068Function calling only with none · Chat Completions+3
CheaperClaude Haiku 5.5xhigh+24%-87%$0.12Remove sampling parameters+5
CheaperGemini 3 Flash Previewhigh+19%-27%$0.50 / $3
Kimi K2.6unstatedOpen+20%+10%$0.95 / $4Memory (4-bit): 452 GB → 616 GB