Vision catalogue · Verified local path

Qwen2.5-VL 3B local guide

Compact multimodal model for images, documents, charts and video understanding.

Choose an app

Start with Recommended. No terminal commands are shown.

Compare vision models
Open model filesRecommended verified starting pointOpen in UnslothLaunches Unsloth Desktop on this modelOpen MLX setupApple Silicon runtimeCheck my machineSee whether this model fits your saved hardwareOpen official sourcePublisher repository and primary evidence

Desktop app links require the app to be installed. If nothing opens, LocalClaw will show app-download and model-file fallbacks.

What it does

Visual Question AnsweringDocument UnderstandingVideo Understanding

Quantized MLX and llama.cpp paths make the 3B checkpoint practical on 16 GB machines.

Hardware figures are practical entry floors, not performance guarantees. Resolution, duration, precision, offloading and runtime versions can materially change memory use.

Strengths

  • Images, documents and video
  • Strong multilingual understanding
  • Several local runtimes

Limits to know

  • Video inputs increase memory use