Our benchmark BenchCAD is featured as a multimodal benchmark in Anthropic’s official Claude Fable 5 & Claude Mythos 5 System Card: a dedicated section (§8.16.4, pp. 282–283) evaluates their frontier models on BenchCAD’s Vision2Code task — with two figures and a Python-tools ablation.

BenchCAD Vision2Code scores in Anthropic's system card From Anthropic’s Claude Fable 5 & Claude Mythos 5 System Card (§8.16.4).