StepFun AI

Step 3.5 Flash is a sparse Mixture-of-Experts model by StepFun with 196.81B total parameters (196B backbone + 0.81B MTP head) and ~11B active per token. Built on a 45-layer transformer with 288 routed experts (Top-8 selection), 3:1 SWA attention ratio, and 256K context. Achieves 100–300 tok/s throughput (peaking at 350 tok/s for coding) for frontier reasoning and agentic tasks.

A 1T parameter multimodal Mixture-of-Experts model optimized for long-horizon coding, agentic tool use, and image/video understanding. Served via NVIDIA NIM with fast inference and structured output support.

Step 5 Preview is StepFun's next-generation flagship base model designed for real-world tasks, with a focus on programming and professional expertise. A multimodal (text, image, video in / text out) base model with a 1M-token context and adjustable reasoning effort.