Step 5 Preview

by StepFun Coding Agentic Reasoning Long Context Multimodal Image Input Video Input

Step 5 Preview is StepFun's flagship model for agentic work, built on a sparse Mixture-of-Experts architecture (27B active / 600B total parameters).

Choose a model to compare against Step 5 Preview

Specifications

Specifications for Step 5 Preview
LabStepFun
Context window1M tokens
Input price $1.00/1M
Output price $2.70/1M
ReleaseSep 2026

Strengths

  • Frontier-class agentic coding at open-tier pricing ($1.00 in / $2.70 out per 1M tokens)
  • 1M-token context window with text, image, and video input
  • Standout finance-domain results (FrontierFinance 66.4%, ahead of open peers)
  • Fast inference for a reasoning model with adjustable reasoning effort (low/medium/high)
  • Native tool calling, structured output, and prompt caching for agent workflows

Weaknesses

  • All launch benchmarks are vendor-reported with no technical report published yet
  • Verbose output inflates output-token costs on high-effort runs
  • Trails closed-frontier models on the hardest long-horizon coding tasks

Best for

  • Agentic coding workflows and long-horizon software engineering tasks
  • Financial analysis and research over large document sets
  • Cost-sensitive long-context applications with repeated prompts

In Depth: Step 5 Preview

Summary

Step 5 Preview is an AI model from StepFun.

Released in Sep 2026. It takes text, image, and video input and produces text output, with a context window of 1M tokens. Input pricing is $1.00 per 1M tokens and output is $2.70 per 1M tokens on OpenRouter.

Sources & Further Reading

Related Models