OCRRouter

Qwen-VL vs Google Cloud Vision OCR

Pick Google Cloud Vision OCR for high-volume plain text extraction from photos, signs, labels and handwritten notes when you want a hosted API with a clear per-page price. Pick Qwen-VL when you need Markdown, HTML or JSON output, want to self-host under Apache-2.0, or want OCR and reasoning in one model. The main split is hosted, plain-text OCR versus an open, general model that can return layout but can loop on bad scans.

Qwen-VLqwen/qwen-vlGoogle Cloud Vision OCRgoogle/cloud-vision
TypeOpen sourceOCR API
Price2 credits / page2 credits / page
Per 1,000 pages$2.00$2.00
Typical speed2.0 s / page1.0 s / page
Best forFlexible extraction with promptsPhotos and simple scans
InputsPDF, imagesPDF, images
Languages30+ languages100+ languages
Max pages per request2002,000
Tables✓ YesNo
Handwriting✓ Yes✓ Yes
Math✓ YesNo
Bounding boxes✓ Yes✓ Yes
ConfidenceNo✓ Yes
FormsNoNo
Multilingual✓ Yes✓ Yes
On OCRRouterAvailableAvailable

Prices are what you pay on OCRRouter. Speed is the provider's typical figure, not a benchmark we ran.

Which to choose

Choose Qwen-VL if

  • You can self-host it under Apache-2.0, in sizes from 2B to 235B-A22B MoE, and pay only for compute.
  • It outputs plain text, HTML, bounding boxes, JSON and Markdown, including layout positions, where Cloud Vision gives no Markdown or layout structure.
  • It handles OCR in 32 languages and mixes OCR with reasoning or VQA in one model, including photos and screenshots with text.

Choose Google Cloud Vision OCR if

  • It is a hosted API with a published price: first 1,000 units per month free, then $1.50 per 1,000, dropping to $0.60 per 1,000 above 5M units.
  • It returns a page, block, paragraph, word and symbol hierarchy with bounding polygons, and supports handwriting through DOCUMENT_TEXT_DETECTION and language hints.
  • It does not loop on poor images the way Qwen-VL reports describe, and async batch takes up to 2,000 image files per request.

Key differences

Qwen-VLGoogle Cloud Vision OCR
LicenceApache-2.0 (Qwen3-VL repository)Proprietary
How you run itSelf-host open weights, or use Alibaba Cloud as a hosted API. FP8 checkpoints need an H100-class GPU and CUDA 12+.Hosted API only, over REST or client libraries on Google Cloud. No on-premise option.
PriceFree self-hosted, you pay for compute. Hosted pricing on Alibaba Cloud Model Studio varies by model.1,000 units per month free, then $1.50 per 1,000 up to 5M, then $0.60 per 1,000. One PDF page is one unit.
Output formatsPlain text, HTML, bounding boxes, JSON, MarkdownJSON, plain text, bounding boxes. No Markdown, so you rebuild reading order yourself.
Tables and layoutOutputs layout positions and a Qwen HTML document formatNo table or layout structure. Google points to Document AI for forms and tables.
BenchmarkQwen 3 VL 8B scores 64.6 on olmOCR-bench in an independent table, below dedicated OCR modelsNot published
Known failureCan loop and repeat the same text, worse with greedy decoding and on low-quality scans. GGUF builds of Qwen2.5-VL-32B often get stuck on scanned documents.Billing must be enabled even for the free 1,000 units. PDF and TIFF need async processing via Cloud Storage.

Sources

What each one is

Qwen-VL

Alibaba's open-weight vision language models (Qwen2.5-VL, Qwen3-VL). Strong general OCR and grounding.

Qwen-VL details

Google Cloud Vision OCR

Google Cloud's image OCR with text and document text detection, word boxes and confidence.

Google Cloud Vision OCR details

Try both on your own file

Run the file once with each model and compare the output.

PDF, PNG, JPG, WebP, HEIC. Up to 20 MB each, 10 files.No file?

Use one, fall back to the other

{"model": "qwen/qwen-vl","fallbacks": ["google/cloud-vision"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Qwen-VL alternativesGoogle Cloud Vision OCR alternativesBest OCR models in 2026