South Korea32B parameters128K contextVision + video

Explore HyperCLOVA X SEED Think 32B

Use NAVER's Korean-centered vision-language reasoning model with text, image, and video inputs for analysis and agentic tasks.

Model

naver-hyperclovax/HyperCLOVAX-SEED-Think-32B

Input

Configure your request

System prompt Optional

Set the assistant role and multimodal instructions.

Conversation history Optional

Optional text and multimodal message history as JSON.

Prompt or question

Images or video Optional

Optional image or video inputs for multimodal understanding.

Parameters

Thinking modeEnable the model’s controllable deep-reasoning mode. Stream response Sample output

Temperature

Top P

Top K

Maximum output tokens

Text, visual tokens, and output must fit the 128K context.

Repetition penalty

Seed

Stop sequences

Function tools (JSON)

Optional tool definitions for agentic tasks.

Tool choice