Index leader
Claude Fable 5
60 / 100
Index per dollar
DeepSeek V4 Pro
$0.04 per task
Evaluated field
15 models
Routed models and clearly marked market comparisons
← Lower cost

Each point is one retained measurement; this view does not connect the points into a market frontier.

Higher score ↑
OpenAIAnthropicMoonshot AIGrokMetaGoogleDeepSeekMiniMaxNVIDIAZ.aiArtificial Analysis Intelligence Index v4.1 · Jul 2026

Independent benchmark index across reasoning, knowledge, coding, tool use, and long-context evaluations. Cost is the weighted observed cost per Intelligence Index task. Points are not connected because these retained benchmark configurations are not the source's full effort-level Pareto frontier.

Looking for the full sortable table? The Leaderboard ranks every routed model with public evidence and Core judgment kept separate.