← Back to models

Qwen3-VL

generatorAnswersImage

Answer questions about images and read the text in them with a 2B vision-language model. The pull command selects the official Q4_K_M decoder with its Q8_0 vision projector.

antfly inference pull Qwen/Qwen3-VL-2B-Instruct-GGUF:q4-k-m-bundle-v1
By QwenLicense: apache-2.0Hugging Face source

Technical Specs

Architecture
Qwen3-VL
Parameters
2B
Q4_K_MGGUFmultilingual

Usage

Pull the model

antfly inference pull Qwen/Qwen3-VL-2B-Instruct-GGUF:q4-k-m-bundle-v1

Start Antfly Inference

# With Antfly standalone (Antfly Inference enabled by default)
antfly standalone

# Or run Antfly Inference on its own, listening on 127.0.0.1:8090
antfly inference run

Use in configuration

{
  "generators": {
    "local": {
      "provider": "antfly",
      "model": "Qwen/Qwen3-VL-2B-Instruct-GGUF:q4-k-m-bundle-v1"
    }
  }
}