S

Step 5 Preview

StepFun·1.02M context·Text·Released Sep 15, 2026

Frontier multimodal reasoning with a 1M token context, image and video input, parallel tool calling, and strict JSON schema output.