← All models

Qwen-VL-OCR

Published Publisher Alibabaqwen-vl-ocr-2025-08-28

Overview

Qwen-VL-OCR is an optical character recognition model trained on Qwen-VL. It brings image-text recognition, parsing, and processing tasks together in one model. It is intended for extracting and processing text from images. (help.aliyun.com)

Release

Alibaba’s model-release listing dates qwen-vl-ocr-2025-08-28 to September 13, 2025. The date in the model ID identifies the snapshot; it is not the listing’s publication date. Official release listing. (help.aliyun.com)

Limitations and version status

  • Alibaba’s Qwen-OCR documentation says the November 20, 2025 version and earlier versions process only the latest message and do not retain conversational context. Custom system messages are not supported; instructions must be sent in the user message. Small or low-resolution text can cause hallucinations, and accuracy on questions unrelated to text extraction is not guaranteed. Official usage documentation. (alibabacloud.com)
  • The unversioned qwen-vl-ocr alias was updated from this snapshot to qwen-vl-ocr-2025-11-20 in February 2026; it should not be assumed to identify the August snapshot. Official update notice. (alibabacloud.com)
  • Current availability of this exact snapshot is unverified. It was not listed on the currently retrieved domestic and international pricing pages. (help.aliyun.com)

Pricing

The following are previously published prices, not verified current offers. Both apply to Alibaba Cloud Model Studio in China (Beijing), and input and output tokens are billed separately. The domestic and international-site prices are separate listings, not a currency conversion. Checked November 18, 2026.