Benchmark

ZeroBench-main (w/tools)

Vision benchmark ZeroBench main split, tool-assisted.

Score history →

Model scores

#	Model	Score
1	GPT 5.6-Sol	53%
2	Claude Opus 5	52%
3	DeepSeek V4.1-Flash 1017	51%
4	DeepSeek V4.1-Flash 0910	49%
5	Kimi K3	41%