Step 5 Preview
Stepfun馃嚚馃嚦
stepfun/step-5-previewStep 5 Preview is StepFun's flagship multimodal model, a sparse mixture-of-experts (600B total, about 27B active parameters) with a 1M-token context window and text, image, and video input, released September 20, 2026 with open weights announced to follow. It is aimed at agentic coding, software engineering, and professional knowledge work, with particular strength in finance.
Cost rate
0.6x
Context
1M
Released
Oct 8, 2026
Input
TextImageVideo
Output
Text
Support
ReasoningTool callingStructured outputs
Best at
The categories where this model ranks highest.
Image understanding
#55 路 top 33%
Software & IT Services
#69 路 top 17%
Mathematical
#72 路 top 18%
Life, Physical, & Social Science
#84 路 top 20%
Performance
Median latency and throughput measured across recent requests.
Throughput92 tok/s
Latency