cd /news/large-language-models/step-5-preview-stepfun-s-frontier-mo… · home topics large-language-models article
[ARTICLE · art-135928] src=platform.stepfun.ai ↗ pub= topic=large-language-models verified=true sentiment=· neutral

Step 5 Preview: StepFun's frontier model with 1M context and video input

StepFun released Step 5 Preview, a flagship model for agentic work that natively supports text, image, and video input and offers a 1M-token context window. StepFun says the model delivers frontier-level performance across software engineering and professional knowledge work, with particular strength in finance, and supports multi-step agent tasks, long-document processing, and multimodal understanding. Step 5 Preview is billed on actual input and output token usage and can be integrated via the StepFun Open Platform Chat Completions API or Claude Code with Step Plan.

by read2 min views6 publishedSep 21, 2026
Step 5 Preview: StepFun's frontier model with 1M context and video input
Image: source

Step 5 Preview is StepFun’s flagship model for agentic work. It delivers frontier-level performance across software engineering and professional knowledge work, with particular strength in finance. It natively supports text, image, and video input and provides a

1M-token context window, making it suitable for tasks that require large amounts of information, tool calls, and continuous progress toward a deliverable.

At a glance #

Core capabilities #

Long-context understanding and reasoning #

Analyze long documents, multiple source materials, and conversation history within a 1M-token context.

Typical tasks: cross-document question answering and research material organization.

Programming and software engineering #

Work across multiple programming languages and engineering tasks, with tools to advance development, modification, and verification.

Typical tasks: troubleshooting, code changes, and test creation.

Multi-step Agent tasks #

Use tools provided by an application to retrieve information, process documents, and continue through multi-step work. Typical tasks: deep research and analytical reporting.

Multimodal understanding #

Combine images, video, and text to extract and analyze multimodal information.

Typical tasks: chart analysis, screenshot question answering, and video summarization.

Use cases #

  • Software development : Combine code, dependency information, and error logs to locate problems, modify code, and recommend tests.
  • Long-document processing : Organize multiple sources in a long context, extract and summarize information, and cross-check details.
  • Research and analysis : Work with search, code execution, or document tools to break down tasks and produce structured results.
  • Multimodal understanding : Analyze screenshots, charts, and video, and output conclusions in a specified format.

Input and output #

Image and video input

See the

Quickstartfor image and video request examples. For complete limits, see

[Image understanding best practices](https://platform.stepfun.ai/docs/en/guides/developer/image-chat)and

[Video understanding best practices](https://platform.stepfun.ai/docs/en/guides/developer/video-chat).

Features and configuration #

Integration #

  1. Create an API key on the StepFun Open Platform.

  2. Open the Quickstart , select Step 5 Preview, and switch between the text chat, image understanding, and video understanding examples.

  3. Send a request with cURL or Python and read the result from choices[0].message.content .

  4. Before going live, use the API reference to add timeout, retry, error-handling, and key-management logic.

Quickstart #

See text, image, and video request examples.

Chat Completions API #

See the complete request parameters, response fields, and usage information.

Claude Code integration #

Configure Step Plan and enable 1M context.

Pricing and rate limits #

Step 5 Preview is billed based on actual input and output token usage. For current prices, cache billing rules, and account rate limits, see

Pricing and Rate Limits.

More information #

[Quickstart](https://platform.stepfun.ai/docs/en/quickstart/overview)·

[API reference](https://platform.stepfun.ai/docs/en/api-reference/chat/chat-completion-create)·

[Claude Code integration](https://platform.stepfun.ai/docs/en/step-plan/integrations/claude-code)·

[Pricing and Rate Limits](https://platform.stepfun.ai/docs/en/guides/pricing/details)
── more in #large-language-models 4 stories · sorted by recency
── more on @stepfun 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/step-5-preview-stepf…] indexed:0 read:2min 2026-09-21 ·