Accuracy Distribution
Success rate based on wordbank.
01
GPT-5.4 Mini (high)
77%
02
Gemini 3 Flash (dynamic)
75%
03
GPT-5 Mini (medium)
71%
04
Gemini 3.1 Flash Lite (medium)
62%
05
GPT-5.4 Mini (medium)
60%
06
Gemini 3.1 Flash Lite (minimal)
46%
07
GPT-5.4 Nano (medium)
44%
08
Claude Haiku 4.5
33%