> ## Documentation Index
> Fetch the complete documentation index at: https://platform.stepfun.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Reasoning Models

## Model overview

Reasoning models are designed for deep analytical work, including logic, mathematics, coding, and other complex tasks.

## Models

Reasoning / multimodal

Step 5 PreviewFlagship model for agentic work

Context (tokens)1M

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 can combine tools with documents to complete multi-step tasks.

Coding
Knowledge work
Agent
Image and video understanding

View model overview 

[View the Step 5 Preview model overview](https://platform.stepfun.ai/docs/en/guides/models/step-5-preview.md)

Reasoning / multimodal

Step 3.7 FlashHigh-speed multimodal reasoning

Context (tokens)256K

Combines text, image, and video input with high-speed reasoning and tool calling for Agent, coding, and multimodal understanding workloads.

Reasoning
Agent
Image understanding
Video understanding

View model overview 

[View the Step 3.7 Flash model overview](https://platform.stepfun.ai/docs/en/guides/models/step-3.7-flash.md)

Reasoning / text

Step 3.5 FlashHigh-speed language reasoning

Context (tokens)256K

Designed for logic, mathematics, software engineering, and deep research in text-based reasoning and Agent workflows.

Reasoning
Coding
Agent
Deep research

View model overview 

[View the Step 3.5 Flash model overview](https://platform.stepfun.ai/docs/en/guides/models/step-3.5-flash.md)

A model's context length limits the total input and output it can process in one request. See each model overview for its exact rules.

## Get started with reasoning models

### Developer guides

> [card] [Reasoning model guide](https://platform.stepfun.ai/docs/en/guides/developer/reasoning.md)
>
> Learn recommended patterns for complex tasks, tool calling, and long-context work.

> [card] [Migrate from OpenAI](https://platform.stepfun.ai/docs/en/guides/developer/openai.md)
>
> Move an OpenAI-compatible integration to StepFun with minimal changes.

> [card] [Build multi-turn conversations](https://platform.stepfun.ai/docs/en/guides/developer/multiple-round.md)
>
> Manage message history and context for stable multi-turn experiences.

> [card] [Output JSON](https://platform.stepfun.ai/docs/en/guides/developer/json-mode.md)
>
> Return machine-readable structured JSON for application integration.

> [card] [Stream responses](https://platform.stepfun.ai/docs/en/guides/developer/stream.md)
>
> Render output as it arrives to reduce perceived wait time.

> [card] [Use tool calling](https://platform.stepfun.ai/docs/en/guides/developer/tool-call.md)
>
> Let a model call search, databases, or business functions.

> [card] [Prompt caching](https://platform.stepfun.ai/docs/en/guides/developer/prompt-cache.md)
>
> Reuse repeated context to lower cost and improve response time.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.