Qwen3-VL-8B-Instruct

Name: Qwen3-VL-8B-Instruct
Author: Qwen

by Qwen

8B vision-language model with 262K context and strong visual reasoning

2.8Mdl/month

8.77Bparams

HuggingFace Run it on your own data, free

Identifiers

Model ID

Qwen/Qwen3-VL-8B-Instruct

Feature URI

mixpeek://image_extractor@v1/qwen3_vl_8b_v1

Overview

Qwen3-VL-8B-Instruct is Alibaba's instruction-tuned vision-language model that combines an 8B parameter dense language model with a 400M SigLIP-2 vision encoder. It supports text, image, and video understanding with a native 262K token context window extensible to ~1M tokens, delivering performance that surpasses models 3x its size on key benchmarks.

On Mixpeek, Qwen3-VL-8B powers rich visual understanding tasks including scene captioning, document analysis, and video comprehension where you need detailed visual reasoning without the cost of running a 30B+ model.

Architecture

Early-fusion multimodal architecture built on a dense hybrid foundation of Gated Delta Networks and Gated Attention. The 8B LLM backbone is augmented with a 400M SigLIP-2 SO vision encoder, two-layer MLP mergers, and DeepStack adapters for multimodal and video capabilities.

Mixpeek SDK Integration

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so Qwen3-VL-8B-Instruct 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: { "multimodal-embedding": embeddingOfModelOutput },
        },
      ],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// universal_extractor@v1 runs google/gemini-embedding-2
// (3072-d) over a bucket, with no inference of your own.