Skip to content
← Models
Open weightsAlibaba Qwen/10 trustVerified
Qwen2-VL-2B logo

Qwen2-VL-2B

The tiny 2B open vision-language model built for phones and edge devices.

Visit official page →

About

Qwen2-VL-2B is the smallest member of Alibaba's Qwen2-VL family, released in August 2024 under Apache 2.0. It combines a dynamic-resolution ViT vision encoder with a 1.5B-parameter Qwen2 language decoder, delivering image, video, and multilingual document understanding in a package explicitly optimized for potential mobile deployment. Despite the size, it held its own on video understanding and document QA against other models in its weight class.

The honest picture: a 2B VLM is a compromise by definition; it handles straightforward visual questions well but falls behind larger siblings on complex reasoning, dense documents, and long videos. Its real legacy is as an edge-deployment pioneer, with community ports to on-device runtimes like LiteRT-LM proving the concept. For current work, Qwen2.5-VL-3B and Qwen3-VL-2B are the natural successors, but the 2B remains a useful reference for tiny multimodal builds.

Specs

Context window

Pricing

Free quota

API access

Weights

Open weights

Released

August 29, 2024

Benchmarks

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

on-device-visionmobile-aiedge-vqatiny-multimodal

Trust score — 8/10

/10 trustVerified

Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 117; Trending score 80/100 from researcher assessment.

More in the qwen2-vl family

Sources