Standard & Paid Price
| Meter | Standard Price |
|---|---|
| Input price | $1.0000 / 1M tokens |
| Completion price | $2.7000 / 1M tokens |
| Cache read price | $0.0500 / 1M tokens |
Description
StepFun step-5-preview — StepFun's new flagship model for coding and professional knowledge work, released as a preview. 1M-token context with up to 64K output tokens, and native understanding of images and video alongside text. Reasoning is always on: it comes back as reasoning_content on /v1/chat/completions and as thinking blocks on /v1/messages, bills as output, and draws from max_tokens, so a limit of a few hundred tokens can be spent entirely on reasoning and return no text; leave generous headroom. Depth is selectable per request through reasoning_effort (low / medium / high). Supports multi-step tool calling, JSON Mode and JSON Schema structured output, streaming, and prompt caching. Built for long-document analysis, software engineering, and multi-step agent work. On /v1/responses, send the whole conversation in the input array on every turn: no conversation state is kept upstream, so requests that carry previous_response_id or conversation are refused.
Capabilities
- Tasks
- chat, reasoning, vision, file-video-understanding
Decision evidence
Step 5 Preview is StepFun's sparse mixture-of-experts flagship (600B total, 27B active per token) with a 1M-token context window and text, image and video input. It is API-only today; StepFun has announced open weights for 15 October, and the reviewed tests are mixed rather than one-sided.
2026-09-25 · Markdown evidence · JSON data
API endpoints
POST /v1/chat/completions(openai)POST /v1/messages(anthropic)POST /v1/responses(openai-response)