StudyArena

Step3 VL 10B

MultimodalOpen weightsText + images10B parameters

Step family model by StepFun.

About the Step family

StepFun's Step series (Step 3.5/3.7 Flash, Step3 VL vision models) — multimodal reasoning models with cost-efficient attention designs.

All Step models

How this one is configured

This configuration ~10B parameters; multimodal (vision).

Who makes it

StepFunChinaShanghai lab building the multimodal Step model series.

More from Step

← Back to all models