Step 3.7 Flash

LLM

Stepfun

Step 3.7 Flash is a sparse MoE vision-language model built for agentic workflows combining perception, search, and reasoning. It supports a 256k context window and offers three selectable reasoning levels to balance speed and

Context tokens

262,144

Output tokens

8,192

Docs

Model documentation

Released

Jun 12, 2026

Schema

Supports three selectable reasoning levels (low, medium, and high) that let you balance speed and depth.

Schema documentation

Capabilities

Tool use

Structured output

Thinking

Multimodal

Supported input media

image

text

video

Supported tools

No tools enabled.

Pricing

Type	Credits	Units
Input	—	Credits per 1k tokens
Output	—	Credits per 1k tokens

Variants