StepFun's Step 5 Preview model, released on October 8, 2026, is now available on OpenRouter. It is a sparse Mixture-of-Experts model with 27 billion active parameters out of 600 billion total. The model offers a context length of 1 million tokens, enabling it to process very long documents or codebases in a single pass.
This image illustrates the model's debut on OpenRouter, confirming its availability and providing visual context for the announcement.
Designed for agentic work, software engineering, and professional knowledge tasks — with particular strength in finance — the model supports tool calling for function calls and structured outputs via a JSON schema in the response_format parameter. It accepts multimodal input including text, images, and video while generating text output. Performance metrics show a P50 latency of 3.50 seconds, a throughput of 64 tokens per second, and high reliability with 99.99% uptime and 99.68% availability over the last three days.
Pricing on OpenRouter is set at $1.00 per million input tokens, $2.70 per million output tokens, and $0.05 per million tokens for cache read. The model is already seeing real-world adoption, with public apps such as OpenClaw (204B tokens), Hermes Agent (23.7B tokens), Cline (16.7B tokens), Nous Research API (7.75B tokens), and Kilo Code (2.26B tokens) generating significant traffic, indicating its suitability for AI agent and coding assistant workflows.