qwen3-vl:4b

Upstream reference: Qwen3-VL-4B-Instruct

Publisher: Qwen Team, Alibaba Cloud

The Qwen3-VL card describes instruction-tuned vision-language generation, including image/video understanding and structured visual interaction.

Publisher-described context, separate from Sigma5C observations and recorded catalog statistics.

Published upstream specifications

Published native context

256K tokens, publisher notation Source

Upstream license label

Apache-2.0 Source

Checkpoint branch

Instruct Source

Publisher-described uses

• Visual descriptions, document reading and visual tool workflows. Source

Reference limitations

• The Instruct card must not be relabeled as the separate Thinking edition; advertised extensions and enabled tools depend on the application. Source

A recorded tag and an upstream card association do not verify the deployed checkpoint, quantization, configured context or current hosted availability.

Publisher sources

• Qwen3-VL-4B-Instruct official documentation — Qwen Team, Alibaba Cloud. Checked 2026-11-02.

Links

Recorded references do not establish downloadable weights, licensing or current availability.

• Alibaba Cloud Qwen — official site
• Ollama
• Hugging Face

Where qwen3-vl:4b comes from

Named in 2 observation pages

• 3B-7B local models — all 53 · V1
• Models as data workers · V1