Artificial Analysis Intelligence Index
58
Claude Opus 5.5 leads, Oct 20, 2026
- 40 models
- Higher is better
Boards
Every benchmark, one page.
12 boards in 7 categories. We collect the scores each board and lab publishes, link every one to its source, and run no benchmarks ourselves.
12boardsin 7 categories
Overall 2 boards
LMArena text
1525Gemini 4 Argon leads, Oct 8, 2026
- 3 models
- Higher is better
Reasoning and knowledge 3 boards
Humanity's Last Exam
67.7%Claude Opus 5.5 leads, Sep 22, 2026
- 4 models
- Higher is better
- Percent
ARC-AGI-2
95.0%GPT-6 Astra leads, Oct 19, 2026
- 3 models
- Higher is better
- Percent
GPQA Diamond
96.0%GPT-6 Astra leads, Oct 20, 2026
- 1 model
- Higher is better
- Percent
Coding and agents 3 boards
SWE-bench Verified
96.0%Claude Opus 5 leads, Jul 24, 2026
- 1 model
- Higher is better
- Percent
Artificial Analysis Coding Index
82Claude Fable 5.1 leads, Oct 20, 2026
- 23 models
- Higher is better
OSWorld-Verified
86.1%Qwen3.8-Max leads, Aug 3, 2026
- 1 model
- Higher is better
- Percent
Speed 1 board
Output speed
239Claude Haiku 5.5 leads, Oct 20, 2026
- 26 models
- Higher is better
Image 1 board
Text-to-image arena
1222GPT Image 2.5 Sunburst leads, Oct 20, 2026
- 15 models
- Higher is better
Video 1 board
Text-to-video arena
1135Wan 3.0 leads, Oct 20, 2026
- 15 models
- Higher is better
Audio 1 board
Speech arena
1333Eleven v4 Turbo leads, Oct 20, 2026
- 15 models
- Higher is better