AlibabaQwen OCR
Qwen-VL-OCR
Qwen-VL-OCR is based on Qwen3-VL. Its built-in tasks include text localization, table and formula recognition, and image rotation correction.
Technical details
The default output limit is 4,096 tokens. Requests for a larger limit require account-specific approval.Qwen OCR guide
Practical use
Practical guidance. For long pages, check whether the answer ends before all content is returned. Separate transcription errors from field-mapping errors so the right stage can be corrected.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Model identifiers | qwen-vl-ocrqwen-vl-ocr-2025-11-20 |
| Snapshot | qwen-vl-ocr-2025-11-20; earlier 2024 versions are not recommended.Qwen OCR guide |
Related models
- Qwen3.5-OCR
Alibaba ยท Qwen OCR
Qwen3.5-based OCR model with multi-turn conversation and PDF parsing.
Launch date not confirmedAPI