[Models](/models)/[Captioning](/models?category=captioning)/sensenova/SenseNova-U1-8B-MoT

HFScene CaptioningApache 2.0

# SenseNova-U1-8B-MoT

by sensenova

8B any-to-any multimodal model for image understanding, generation, and editing

32.7Kdl/month

281likes

8Bparams

[HuggingFace](https://huggingface.co/sensenova/SenseNova-U1-8B-MoT)[Run it on your own data, free](https://studio.mixpeek.com/namespaces/create?mode=managed&utm_source=homepage&utm_medium=model_page&utm_campaign=mvs_onramp)

Identifiers

Model ID

`sensenova/SenseNova-U1-8B-MoT`

## Deploy SenseNova-U1-8B-MoT

Single-tenant

Mixpeek has no managed extractor for this model. On a single-tenant deployment you upload the weights and a custom plugin serves them next to the rest of your pipeline.

[Read the customer-hosted guide](https://mixpeek.com/docs/customer-hosted/overview) [Talk to us about single-tenant](/demo)

## Overview

SenseNova-U1-8B-MoT is an any-to-any multimodal model tagged for feature extraction, image-to-text, text-to-image, image editing, and custom-code inference. That mix matters for agents because perception is often not a single captioning call: an agent may need to inspect an image, generate an explanation, propose an edit, and preserve evidence of what changed.

On Mixpeek, SenseNova U1 fits pipelines that retrieve visual evidence first, then ask a multimodal model to explain or transform that evidence. It is especially relevant for creative QA, ad review, product imagery, and human-in-the-loop visual analysis.

## Architecture

8B-class mixture-of-transformers style any-to-any multimodal model. Supports image-to-text, text-to-image, image editing, and feature extraction paths according to the Hugging Face model metadata.

## Mixpeek SDK Integration

IngestRetrieve

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so SenseNova-U1-8B-MoT runs
// on your side and the output is upserted through POST
// /v1/namespaces/{namespace\_id}/documents/upsert. On Enterprise the other
// path is to upload the weights instead: POST /v1/namespaces/{id}/models
// accepts the huggingface format and a custom plugin loads them.
const res = await fetch(
  "https://api.mixpeek.com/v1/namespaces/ns\_your\_namespace/documents/upsert",
  {
    method: "POST",
    headers: {
      Authorization: "Bearer API\_KEY",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      collection\_id: "col\_your\_collection",
      documents: \[
        {
          document\_id: "asset-00412",
          // The model produces text, so it lands in payload. Give the
          // collection a text vector index and embed that text to make it
          // searchable rather than only filterable.
          payload: { extracted\_text: modelOutput, source\_key: "archive/2026/asset-00412" },
          vectors: { "image-embedding": embeddingOfModelOutput },
        },
      \],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// image\_extractor@v1 runs google/siglip-base-patch16-224
// (768-d) over a bucket, with no inference of your own.

## Capabilities

*   Any-to-any multimodal interaction across image and text tasks
*   Image-to-text reasoning for visual evidence review
*   Text-to-image and image-editing paths for iterative agent workflows
*   Apache 2.0 licensed model card metadata on Hugging Face

## Use Cases on Mixpeek

Agent review of creative assets and visual ad variants

Multimodal retrieval followed by image-grounded explanation

Visual QA where an agent needs both inspection and transformation

Product imagery triage and annotation

## Performance

Input SizeVariable

GPU LatencyInput dependent

GPU ThroughputBatch dependent

GPU Memory~18 GB

Any-to-any models should be routed to the narrowest task path needed for the agent step.

## Common Pipeline Companions

[

jinaai/jina-embeddings-v5-omni-small-retrieval

Multimodal retrieval layer before generative visual reasoning

](/model/jinaai/jina-embeddings-v5-omni-small-retrieval)[

Qwen/Qwen3-VL-Reranker-2B

Cross-modal reranking for retrieved creative evidence

](/model/Qwen/Qwen3-VL-Reranker-2B)

## Explore on Mixpeek

[

More Captioning models

Compare alternatives in this category

](/models?category=captioning)[

Best Document AI Platforms

Hand-picked tools & platforms compared

](/curated-lists/best-document-ai-platforms)[

OCR & Document AI Internals

Deep-dive technical guide

](/guides/ocr-document-ai-internals)[

Feature Extractors

See how Mixpeek runs models as extractors

](/extractors)[

Mixpeek Vector Store

Store & search embeddings at scale

](/mvs)[

Pricing

Usage-based pricing for pipelines

](/pricing)[

All Curated Lists

Compare models, APIs & infrastructure

](/curated-lists)

### Specification

FrameworkHF

Organizationsensenova

FeatureScene Captioning

Outputtext

Modalitiesvideo, image

RetrieverSemantic Search

Parameters8B

LicenseApache 2.0

Downloads/mo32.7K

Likes281

[

### Research Paper

SenseNova-U1

arxiv.org](https://arxiv.org/abs/2605.12500)

### Build a pipeline with SenseNova-U1-8B-MoT

Add this model to a processing pipeline alongside other extractors. Combine with retrieval stages for end-to-end search.

[Run it on your own data, free](https://studio.mixpeek.com/namespaces/create?mode=managed&utm_source=homepage&utm_medium=model_page&utm_campaign=mvs_onramp)

### Alternative Models

[

Salesforce/blip2-opt-2.7b

Scene Captioning

](/model/Salesforce/blip2-opt-2.7b)[

microsoft/Florence-2-large

Scene Captioning

](/model/microsoft/Florence-2-large)[

google/paligemma2-3b-mix-448

Scene Captioning

](/model/google/paligemma2-3b-mix-448)[

Qwen/Qwen3-VL-8B-Instruct

Scene Captioning

](/model/Qwen/Qwen3-VL-8B-Instruct)