ARC-AGI-2
A hard abstract-reasoning benchmark from ARC Prize focused on novel task generalization.
Benchmarks
Human-readable benchmark definitions with score direction, limitations, and source records.
Humanity's Last Exam (HLE)
A broad expert-level benchmark intended to test difficult questions across many domains and formats.
FrontierMath
An advanced mathematics benchmark curated around difficult research-style math problems.
GPQA / GPQA Diamond
A graduate-level, Google-proof question-answering benchmark with a commonly used Diamond subset.
MMLU-Pro
A harder MMLU-style benchmark for broad knowledge and reasoning.
SWE-bench Verified
A curated SWE-bench split for resolving real software issues with agentic coding systems.
SWE-bench Pro
A professional software-engineering agent benchmark with public and private leaderboard splits.
Terminal-Bench 2.0
A terminal-environment agent benchmark for command-line task completion.
MLE-bench
A machine-learning engineering benchmark based on Kaggle-style competition tasks.
BrowseComp
A browsing benchmark for hard-to-find information and deep web research tasks.
tau-bench / tau2-bench
A tool-use benchmark family for business-process and customer-service style agent tasks.
OSWorld / OSWorld-Verified
A computer-use benchmark for operating-system and GUI agents.
WebArena
A web-agent benchmark for completing tasks in realistic self-hosted web environments.
VisualWebArena
A visual web-agent benchmark extending WebArena with image-grounded web tasks.
LongBench v2
A long-context understanding benchmark with a project leaderboard.
RULER
A synthetic long-context stress benchmark from NVIDIA.
HELMET
A long-context benchmark suite focused on practical application-style tasks.
MMMU-Pro
A harder multimodal academic benchmark derived from MMMU-style tasks.
Video-MME
A video understanding benchmark for multimodal models.
MathVista
A visual math reasoning benchmark for multimodal models.
CharXiv
A chart and figure understanding benchmark for scientific visual reasoning.
SimpleQA Verified
A factuality benchmark for short-answer factual knowledge questions.
FRAMES
A RAG and factual multi-hop question answering benchmark.
GraphWalks
A graph reasoning benchmark candidate that still needs a stable authoritative source before launch.
TerminalWorld-Verified
A terminal-agent benchmark candidate that needs source confirmation.
Artificial Analysis Intelligence Index v4.3.2
Independent 10-evaluation Intelligence Index v4.3.2; scores are not percentages.
AA Terminal-Bench 4.0
Terminal-Bench 4.0 in the Artificial Analysis evaluation harness.
AA Humanity's Last Exam
Humanity's Last Exam as evaluated by Artificial Analysis; the source configuration is preserved.
GDPval-AA v2.1
GDPval-AA v2.1 professional knowledge-work Elo. This is a task lens, not an assessment of every research skill.
AA GDP.pdf
GDP.pdf professional document understanding, including visual tables, charts and diagrams; not an audio/video score.
AutomationBench-AA
AutomationBench-AA multi-step business workflows in the Artificial Analysis harness.
Five-axis capability index
Equal-weight average of five normalized capability indices: coding, reasoning, research, multimodal and agentic.
CursorBench 4.0
Version 4.0 coding-agent evaluation; distinct from CursorBench 3.2.0.
GDPval-AA v2.1
Version 2.1 occupational knowledge-work Elo; distinct from v2.
AA-Briefcase v1.1
Long-horizon knowledge-work evaluation; reported Elo, not a percentage.
OSWorld 2.1 Partial
OSWorld 2.1 partial-credit computer-use evaluation; separate task version from 2.0.
OSWorld 2.0 Offline Partial
Partial reward on the offline v2026.08.08 task set; not latency-simulated OSWorld 2.0.
Chartography with Tools
Visual chart understanding with tools; separate from no-tools results.
Chartography without Tools
Visual chart understanding without tools; separate from tool-enabled results.
CWE-bench v1
Vulnerability-remediation benchmark version 1; distinct from version 0.
Artificial Analysis Intelligence Index
Third-party composite intelligence index; each result must retain the model's reasoning configuration and index version.
FrontierCode 1.1 Main
FrontierCode 1.1 Main result records from the supplied 2026 frontier model reports.
DeepSWE v1.1
DeepSWE v1.1 result records from the supplied 2026 frontier model reports.
Code Arena WebDev
Code Arena WebDev result records from the supplied 2026 frontier model reports.
Terminal-Bench 2.1
Terminal-Bench 2.1 result records from the supplied 2026 frontier model reports.
Terminal-Bench 3.0
Terminal-Bench 3.0 result records from the supplied 2026 frontier model reports.
Terminal-Bench 4.0
Terminal-Bench 4.0 result records from the supplied 2026 frontier model reports.
Terminal-Bench-Science 0.1
Terminal-Bench-Science 0.1 result records from the supplied 2026 frontier model reports.
AutomationBench
AutomationBench result with version or harness unspecified in the supplied source; keep that limitation in result notes.
AutomationBench v1.0.6
Versioned AutomationBench result; preserve harness notes from the supplied Z.ai report.
GDPval-AA v2
GDPval-AA v2 result records from the supplied 2026 frontier model reports.
GDP.pdf
GDP.pdf result records from the supplied 2026 frontier model reports.
CharXiv Reasoning, no tools
CharXiv Reasoning, no tools result records from the supplied 2026 frontier model reports.
CharXiv Reasoning + Tools
CharXiv Reasoning + Tools result records from the supplied 2026 frontier model reports.
LVBench
LVBench result records from the supplied 2026 frontier model reports.
GDM-MRCR v2
GDM-MRCR v2 result records from the supplied 2026 frontier model reports.
OSWorld 2.0
OSWorld 2.0 result records from the supplied 2026 frontier model reports.
OSWorld 2.0 Partial
OSWorld 2.0 Partial result records from the supplied 2026 frontier model reports.
OSWorld 2.0 Strict
OSWorld 2.0 Strict result records from the supplied 2026 frontier model reports.
Agents' Last Exam
Agents' Last Exam result records from the supplied 2026 frontier model reports.
HLE-Verified
HLE-Verified result records from the supplied 2026 frontier model reports.
Humanity's Last Exam with Tools
Humanity's Last Exam with Tools result records from the supplied 2026 frontier model reports.
BioMysteryBench, human-solvable
BioMysteryBench, human-solvable result records from the supplied 2026 frontier model reports.
BioMysteryBench, human-difficult
BioMysteryBench, human-difficult result records from the supplied 2026 frontier model reports.
LABBench2
LABBench2 result records from the supplied 2026 frontier model reports.
NL2Repo
NL2Repo result records from the supplied 2026 frontier model reports.
Toolathlon Verified
Toolathlon Verified result records from the supplied 2026 frontier model reports.
OfficeQA Pro
OfficeQA Pro result records from the supplied 2026 frontier model reports.
Chartography + Tools
Chartography + Tools result records from the supplied 2026 frontier model reports.
BabyVision
BabyVision result records from the supplied 2026 frontier model reports.
MVBench
MVBench result records from the supplied 2026 frontier model reports.
MMVU
MMVU result records from the supplied 2026 frontier model reports.
CursorBench 3.2.0
CursorBench 3.2.0 result records from the supplied 2026 frontier model reports.
Vals Finance Agent v2
Vals Finance Agent v2 result records from the supplied 2026 frontier model reports.
Harvey Legal Agent Benchmark
Harvey Legal Agent Benchmark result records from the supplied 2026 frontier model reports.
CharXiv RQ
CharXiv RQ result records from the supplied 2026 frontier model reports.
FrontierCode 1.1 Extended
FrontierCode 1.1 Extended result records from the supplied 2026 frontier model reports.
Database Migration Tasks (Internal)
Database Migration Tasks (Internal) result records from the supplied 2026 frontier model reports.
Artificial Analysis Coding Agent Index v1.4
Artificial Analysis Coding Agent Index v1.4 result records from the supplied 2026 frontier model reports.
FrontierMath Tier 4 (v2)
FrontierMath Tier 4 (v2) result records from the supplied 2026 frontier model reports.
ARC-AGI-3
ARC-AGI-3 result records from the supplied 2026 frontier model reports.
ExploitBench
ExploitBench result records from the supplied 2026 frontier model reports.
ExploitGym
ExploitGym result records from the supplied 2026 frontier model reports.
SRE-Bench
SRE-Bench result records from the supplied 2026 frontier model reports.
ScreenSpot-Pro
ScreenSpot-Pro result records from the supplied 2026 frontier model reports.
BenchCAD
BenchCAD result records from the supplied 2026 frontier model reports.
Mind2Web
Mind2Web result records from the supplied 2026 frontier model reports.
SWE-bench Multilingual
SWE-bench Multilingual result records from the supplied 2026 frontier model reports.
CoWorkBench
CoWorkBench result records from the supplied 2026 frontier model reports.
JobBench
JobBench result records from the supplied 2026 frontier model reports.
LiveCodeBench v6
LiveCodeBench v6 result records from the supplied 2026 frontier model reports.
IFBench
IFBench result records from the supplied 2026 frontier model reports.
ClawEval-MM
ClawEval-MM result records from the supplied 2026 frontier model reports.
RecreationBench
RecreationBench result records from the supplied 2026 frontier model reports.
AndroidWorld
AndroidWorld result records from the supplied 2026 frontier model reports.
Vision2Web
Vision2Web result records from the supplied 2026 frontier model reports.
ERQA
ERQA result records from the supplied 2026 frontier model reports.
RealWorldQA
RealWorldQA result records from the supplied 2026 frontier model reports.
MathVision
MathVision result records from the supplied 2026 frontier model reports.
OSWorld 2.0 Binary
OSWorld 2.0 Binary result records from the supplied 2026 frontier model reports.