AI benchmarks
Model scores as they were reported — verbatim, with test conditions and whether the number is vendor-reported or independently measured. Every result is extracted from sourced event coverage; each benchmark keeps its own leaderboard.
100 meters sprint (World Humanoid Robot Games)
1 result · 1 model
Agentic AI with cross-department orchestration (Talkdesk research)
1 result · 1 model
Agents on Rails
2 results · 2 models
AI coding agents benchmarking (token use per solved task)
2 results · 2 models
AI economy annualized revenue
1 result · 1 model
AI evaluation protocols contest (diplomatic/national-security decision support)
1 result · 1 model
AI productivity (executive survey)
1 result · 1 model
AI spend per employee (top 1% of firms)
1 result · 1 model
AMD datacenter revenue growth (reported)
1 result · 1 model
AMD datacenter revenue (reported)
1 result · 1 model
Annual cost
1 result · 1 model
API pricing
1 result · 1 model
AppWorld
4 results · 1 model
Arc AGI
1 result · 1 model
ARC-AGI-1
2 results · 1 model
ARC-AGI-3
2 results · 1 model
Artificial Analysis Intelligence Index
2 results · 2 models
AtCoder Heuristic Competition
1 result · 1 model
AtCoder Heuristic Contest 058
1 result · 1 model
CommerceAgentBench
1 result · 1 model
Contact center AI deployment coverage (Talkdesk research)
1 result · 1 model
Cyber and biological tasks
2 results · 2 models
DeepSWE
12 results · 6 models
FrontierScience
1 result · 1 model
GPU neocloud providers (published pricing, contracted power)
1 result · 1 model
Hyper-τ-bench
1 result · 1 model
Inference cost
1 result · 1 model
KernelBench-Mega
2 results · 2 models
Legal task accuracy
1 result · 1 model
long-horizon Minecraft evaluation
1 result · 1 model
PitchBook Global VC Ecosystem Rankings (fourth annual report)
1 result · 1 model
Predicted biological age reduction (proteomic aging clocks)
2 results · 1 model
Real-SWE
1 result · 1 model
Remote Labor Index
4 results · 3 models
SWE-Bench Pro
1 result · 1 model
Trailing twelve-month AI economy revenue
1 result · 1 model
Scores are only comparable within one benchmark under matching conditions — we never aggregate across benchmarks into a single ranking.