Key Info

StepFun's Step 5 Preview is described as its largest open-weights release: a 600B MoE model built for agent tasks lasting up to 24 hours, running at nearly 8K tokens/sec/GPU on NVIDIA GB300 using NVFP4.

Highlights

  • 600B MoE open-weights model, described as StepFun's largest open-weights release
  • Built to handle agent tasks lasting up to 24 hours
  • Runs at nearly 8K tokens/sec/GPU on NVIDIA GB300 using NVFP4
  • An NVFP4 checkpoint is available
  • Can be deployed with SGLang