Unlimited OCR vs InternVL3.5
Pick Unlimited OCR to parse long multi-page documents such as reports and books in one request, with Markdown and bounding boxes out. Pick InternVL3.5 for document question answering or a multimodal pipeline where you want one model family in sizes from 1B to 241B. The main split is a dedicated layout-aware parser versus a general vision language model.
Unlimited OCRbaidu/unlimited-ocr | InternVL3.5opengvlab/internvl | |
|---|---|---|
| Type | Open source | Vision model |
| Price | 2 credits / page | 2 credits / page |
| Per 1,000 pages | $2.00 | $2.00 |
| Typical speed | 1.2 s / page | 3.0 s / page |
| Best for | Books and very long documents | Self-hosted document understanding |
| Inputs | PDF, images | images |
| Languages | 30+ languages | Multilingual |
| Max pages per request | 1,000 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | No |
| Math | ✓ Yes | ✓ 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 Unlimited OCR if
- It parses 40+ pages in a single forward pass with roughly constant memory, so long reports and books go in as one request.
- It returns block types and bounding boxes alongside text, which InternVL3.5 does not do by default.
- It has a published benchmark: Unlimited OCR 2 scores 77.1 on MarathonDocBench against 33.7 for the next model, per Baidu.
Choose InternVL3.5 if
- It comes in nine sizes from 1B to 241B-A28B, so you can trade accuracy for cost.
- Models up to 30B run on a single A100, and the 20B-A4B and 30B-A3B MoE variants activate few parameters per token.
- It suits document question answering and general multimodal work, and it can return JSON as well as Markdown and plain text.
Key differences
| Unlimited OCR | InternVL3.5 | |
|---|---|---|
| Licence | MIT | Apache-2.0 |
| How you run it | Self-host with vLLM, SGLang or transformers; hosted endpoint in Baidu Cloud's OCR product | Self-host with Transformers, LMDeploy or vLLM |
| Model size | 3B total, about 500M active per token; Unlimited OCR 2 is 3.34B | 1B to 241B-A28B; the 8B variant is 8.5B |
| Hardware | NVIDIA GPU with CUDA 12.9 or 13.0; Unlimited-OCR-2 under SGLang runs out of memory on 12 GB GPUs | Up to 30B on a single A100; 38B needs two A100s; 241B-A28B needs eight A100s |
| Output | Markdown, plain text, bounding boxes | Plain text, Markdown, JSON; no bounding boxes or layout output by default |
| OCR benchmark | ParseBench mean 46.17; MPDocBench 84.6 and MarathonDocBench 77.1 for Unlimited OCR 2 (author-reported) | Not published as text; the model card shows OCR scores only as chart images |
| Known failure | Text repetition and looping on some pages, reported hallucinated text, and a 32,768 token context cap; text formatting score on ParseBench is only 0.97 | Largest sizes need multiple A100s; not suited to jobs needing word coordinates |
Sources
- Hugging Face model card
- GitHub repo
- MarkTechPost on Unlimited OCR 2
- Issue #55 repetition
- Hugging Face model card (8B)
- InternVL GitHub
What each one is
Unlimited OCR
Open-source model from Baidu that reads hundreds of pages in one pass with flat memory use.
Unlimited OCR 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": "baidu/unlimited-ocr","fallbacks": ["opengvlab/internvl"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Unlimited OCR alternativesInternVL3.5 alternativesBest OCR models in 2026