Skip to content
Loading page…
Loading benchmarks…
Benchmarks
Rankings
Studies
#
↑
Model
Samples
Benchmarks
Rankings
Studies
Web development
Agents
Text
Images
Video
Design
Arena
Sep 9, 2026
43 models
Agent performance · Sep 9, 2026
#
↑
Model
Net improvement
Samples
41
M
Mistral Medium 3.5
Mistral
-10.12%
±1.47%
9,879
42
Minimax M2.7
MiniMax
-13.22%
±1.22%
37,514
43
Gemini 3.5 Flash Lite
Google
-13.73%
±1.45%
23,216
2%
11%
20%
Claude Fable 5.1 (Max): 13.85%
Claude Fable 5.1 (Max)
13.85%
GPT 6 Astra (Max): 12.39%
GPT 6 Astra (Max)
12.39%
Claude Opus 5 (High): 11.06%
Claude Opus 5 (High)
11.06%
Claude Opus 5 (Max): 10.8%
Claude Opus 5 (Max)
10.8%
Claude Fable 5 (High): 9.09%
Claude Fable 5 (High)
9.09%
Claude Opus 4.8 (High): 7.57%
Claude Opus 4.8 (High)
7.57%
Dots: score · lines: published uncertainty