Kraken vs InternVL3.5
Pick Kraken for historical manuscripts, early prints and right-to-left scripts, where you need ALTO or PageXML and are willing to train on your own material. Pick InternVL3.5 for document question answering and self-hosted multimodal pipelines, where you want plain text, Markdown or JSON and a model size that fits your GPUs. The main split is a trainable layout and line OCR tool with coordinates versus a general vision language model without bounding boxes by default.
Krakenkraken/kraken | InternVL3.5opengvlab/internvl | |
|---|---|---|
| Type | Open source | Vision model |
| Price | 1 credit / page | 2 credits / page |
| Per 1,000 pages | $1.00 | $2.00 |
| Typical speed | 1.5 s / page | 3.0 s / page |
| Best for | Historical documents and manuscripts | Self-hosted document understanding |
| Inputs | images | images |
| Languages | Many scripts, incl. right-to-left | Multilingual |
| Max pages per request | 500 | 500 |
| Tables | No | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | No | ✓ Yes |
| Bounding boxes | ✓ Yes | No |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | ✓ Yes | ✓ 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 Kraken if
- It is built for historical documents and non-Latin scripts, with right-to-left, BiDi and top-to-bottom support.
- It outputs ALTO, PageXML, abbyyXML and hOCR plus bounding boxes, which digital humanities tools use.
- Layout analysis, reading order and character recognition are all trainable, and a public Zenodo repository holds community-trained models.
Choose InternVL3.5 if
- It comes in nine sizes from 1B to 241B, so you can trade accuracy for cost.
- Models up to 30B run on a single A100, and the MoE variants 20B-A4B and 30B-A3B activate few parameters per token.
- It suits document question answering and returns plain text, Markdown or JSON.
Key differences
| Kraken | InternVL3.5 | |
|---|---|---|
| What it is | OCR system for historical and non-Latin script material, with trainable layout analysis and recognition | Vision language model family from OpenGVLab, 1B to 241B-A28B parameters |
| Licence and price | Apache-2.0, free and self-hosted (you pay for compute) | Apache-2.0, free and self-hosted (you pay for compute) |
| How you run it | Python library installed with pip on Linux or macOS, x64 or ARM, Python 3.10 to 3.13. No Windows support. | Transformers, LMDeploy or vLLM. Up to 30B on one A100, 38B on two A100s, 241B-A28B on eight A100s. |
| Output formats | ALTO, PageXML, hOCR, plain text, bounding boxes | Plain text, Markdown, JSON. No bounding boxes or layout output by default. |
| Out-of-the-box accuracy | Usually needs training on your own material for good accuracy | OCR scores are published only as chart images, not text values |
| Best document types | Historical manuscripts and early prints; Arabic, Hebrew and other RTL scripts. Not for modern business documents. | Document question answering and self-hosted multimodal pipelines. Not for jobs needing word coordinates. |
| Known problems | Install dependency conflict (issue #816); learned line order lost in segmentation and text output (issue #820) | Larger sizes need multiple A100s; no CPU-only deployment of larger sizes |
Sources
What each one is
Kraken
OCR engine for historical documents and manuscripts, with trainable models for unusual scripts.
Kraken 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": "kraken/kraken","fallbacks": ["opengvlab/internvl"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Kraken alternativesInternVL3.5 alternativesBest OCR models in 2026