Qwen2-VL 72B Instruct

Conditional

Alibaba · Qwen · Released 2024-08

72BMultimodal

The flagship Qwen2-VL tier, competitive with closed multimodal models on visual reasoning, document understanding, and video benchmarks.

Strengths

  • +Top-tier open multimodal benchmark results at release
  • +Dynamic resolution and video understanding
  • +Strong document/OCR and chart understanding

Limitations

  • -Custom Qwen license, not a standard OSI permissive license
  • -72B parameters require multi-GPU serving

License

Qwen License

free with acceptable-use or scale conditions - read it

Hardware

needs multi-GPU or datacenter hardware to run well

Links

Stats

Qwen/Qwen2-VL-72B-Instruct— downloads·— likes

via Hugging Face

Related models