Step3 is a cutting-edge multimodal reasoning model—built on a Mixture-of-Experts architecture with 321B total parameters and 38B active. It is designed end-to-end to minimize decoding costs while delivering top-tier performance in vision–language reasoning. Through the co-design of Multi-Matrix Factorization Attention (MFA) and Attention-FFN Disaggregation (AFD), Step3 maintains exceptional efficiency across both flagship and low-end accelerators.
by Stepfun-ai|66K context|$0.57/M input tokens|$1.42/M output tokens
Endpoints
Available providers for this model, with details on pricing, context limits, and real-time health metrics.
