ChinaAny-to-anyImage + textOpen weights

Explore SenseNova U1

Understand, generate, and edit images with SenseTime's unified any-to-any model, including visual Q&A and interleaved text-image output.

Model

sensenova/SenseNova-U1-8B-MoT

Input

Configure your request

Task

System message Optional

Override the system instruction for interleaved generation.

Prompt or question

Input images Optional

Required for editing and visual Q&A; repeatable for interleaved generation.

Parameters

Generation settings

Thinking modeReason before producing the final text or image. Prompt enhancementExpand short image prompts before generation. Sample text outputUsed for visual Q&A; otherwise greedy decoding.

Output width

Explicit image width in pixels.

Output height

Explicit image height in pixels.

Target pixels

Editing fallback when explicit dimensions are omitted.

Input max pixels

Use “auto” or a numeric image pixel budget.

Resize input imagesFit inputs within the image pixel budget.

Text CFG scale

Controls adherence to the text instruction.

Image CFG scale

Controls adherence to reference images.

CFG normalization

CFG interval start

CFG interval end

Timestep shift

Diffusion steps

Batch size

Seed

Maximum text tokens

Used for visual Q&A.

Text temperature

Text Top P

Text Top K

Repetition penalty

Explore SenseNova U1

Configure your request

Response