Amazon Textract vs InternVL3.5
Pick Amazon Textract for English forms, invoices and large scanned PDF batches on AWS, where you want word and line coordinates from a managed API. Pick InternVL3.5 if you self-host and need document question answering or Markdown and JSON output, or if you need non-Latin languages that Textract does not cover. The main split is a priced, hosted API with layout geometry versus a free Apache-2.0 model family where you pay for GPUs.
Amazon Textractaws/textract | InternVL3.5opengvlab/internvl | |
|---|---|---|
| Type | OCR API | Vision model |
| Price | 25 credits / page | 2 credits / page |
| Per 1,000 pages | $25.00 | $2.00 |
| Typical speed | 1.2 s / page | 3.0 s / page |
| Best for | Forms and key-value extraction | Self-hosted document understanding |
| Inputs | PDF, images | images |
| Languages | 6 languages | Multilingual |
| Max pages per request | 3,000 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | No | ✓ Yes |
| Bounding boxes | ✓ Yes | No |
| Confidence | ✓ Yes | No |
| Forms | ✓ Yes | No |
| Multilingual | No | ✓ Yes |
| On OCRRouter | Available | Available |
Prices are what you pay on OCRRouter. Speed is the provider's typical figure, not a benchmark we ran.
Which to choose
Choose Amazon Textract if
- It returns word and line geometry with confidence scores, so you get bounding boxes without extra work.
- The async API handles PDFs up to 500 MB and 3,000 pages, which suits large scanned batches.
- Basic text detection costs $1.50 per 1,000 pages, and there are no GPUs to run.
Choose InternVL3.5 if
- It is Apache-2.0 and free to self-host, so you pay only for compute and keep documents on your own hardware.
- Nine sizes from 1B to 241B-A28B let you trade accuracy for cost. Models up to 30B fit on a single A100.
- It outputs plain text, Markdown and JSON, which fits document question answering and self-hosted multimodal pipelines.
Key differences
| Amazon Textract | InternVL3.5 | |
|---|---|---|
| How you run it | Hosted AWS API | Self-host with Transformers, LMDeploy or vLLM |
| Licence | Proprietary | Apache-2.0 |
| Price | First 1M pages/month in US West (Oregon): $0.0015/page text detection, $0.015 Tables, $0.05 Forms, $0.015 Queries | Free to download; you pay for compute |
| Output formats | JSON, plain text, bounding boxes | Plain text, Markdown, JSON. No bounding boxes or layout output by default |
| Languages | Text detection covers English, French, German, Italian, Portuguese and Spanish. Handwriting and Queries are English only. Vertical Japanese and Chinese are not supported | Not published |
| Benchmark | 80.9% on RD-TableBench table accuracy (independent) | Not published as text values. OCR scores appear only as chart images on the model card |
| Limits and hardware | Async: PDFs up to 500 MB and 3,000 pages. Sync API takes one page | Up to 30B on one A100, 38B needs two A100s, 241B-A28B needs eight A100s |
Sources
- Textract pricing
- Textract quotas
- Textract hard limits
- Reducto RD-TableBench
- Hugging Face model card (8B)
- InternVL GitHub
What each one is
Amazon Textract
AWS OCR with form key-value pairs, tables and per-word confidence scores.
Amazon Textract detailsInternVL3.5
Shanghai AI Lab's open vision-language family (1B to 241B, August 2025), strong on OCR and document benchmarks.
InternVL3.5 detailsTry both on your own file
Run the file once with each model and compare the output.
Use one, fall back to the other
{"model": "aws/textract","fallbacks": ["opengvlab/internvl"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Amazon Textract alternativesInternVL3.5 alternativesBest OCR models in 2026