Qwen2.5-VL models

Strongest: Qwen2.5-VL-72B-Instruct

Alibaba has since superseded it with Qwen3-VL-235B-A22B (2025-09-22, not in this catalog), and most Qwen3.5 and later open models, from Qwen3.5-9B to Qwen3.5-397B-A17B, are natively vision-language; Model Studio no longer serves the Qwen2.5 series, but the open weights remain on Hugging Face. Qwen2.5-VL-72B-Instruct was the largest open-weight vision-language model from Alibaba when it shipped on 2025-01-26. It accepts interleaved image, video, and text and produces text, with document understanding (tables, forms, charts, OCR) competitive with closed-source frontier VLMs and standout multilingual document parsing. The buyer's sentence: a proven open-weight VLM for document AI and Asian-market visual workloads, at roughly 1/10th the per-token cost of GPT-4o Vision.