GPT vs Unlimited OCR
Pick GPT when you need handwriting, custom JSON extraction with reasoning, or question answering over a document, and you accept a hosted API with per-token cost. Pick Unlimited OCR when you want to parse long PDFs in one pass, self-host under an MIT licence, and get bounding boxes. The main split is hosted and proprietary versus self-hosted and open, with GPT stronger on messy input and Unlimited OCR stronger on long, layout-aware bulk parsing.
GPTopenai/gpt | Unlimited OCRbaidu/unlimited-ocr | |
|---|---|---|
| Type | Vision model | Open source |
| Price | 11 credits / page | 2 credits / page |
| Per 1,000 pages | $11.00 | $2.00 |
| Typical speed | 2.4 s / page | 1.2 s / page |
| Best for | Messy scans, custom instructions | Books and very long documents |
| Inputs | PDF, images | PDF, images |
| Languages | 50+ languages | 30+ languages |
| Max pages per request | 100 | 1,000 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | No | ✓ Yes |
| 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 GPT if
- It suits handwritten notes: an HN user reported about 5% errors on messy cursive versus about 30% for Mistral OCR.
- You can prompt it to extract straight into custom JSON, and it can reason over the document or answer questions about it.
- There is nothing to host: you send files up to 50 MB per request to the API and need no GPU.
Choose Unlimited OCR if
- It parses 40+ pages in a single forward pass with roughly constant memory, so long reports and books go in one request.
- The MIT licence allows commercial use with no copyleft, and self-hosting costs only your compute.
- It returns block types and bounding boxes alongside text, which GPT's standard response does not.
Key differences
| GPT | Unlimited OCR | |
|---|---|---|
| Licence | Proprietary | MIT |
| How you run it | Hosted API only | Self-host with vLLM, SGLang or transformers, or use the hosted option in Baidu Cloud's OCR product |
| Price | Token-based per model; page images add input tokens, so cost per page is not fixed | Free to self-host (you pay for compute) |
| Output formats | Markdown, JSON, plain text; no bounding boxes or confidence scores in the standard response | Markdown, plain text, bounding boxes |
| Long documents | Files up to 50 MB per request | 40+ pages in one pass; context capped at 32,768 tokens, and multi-page input only works in base mode |
| OmniDocBench | GPT-5.2 scored 85.50 on v1.5 and 86.59 on v1.6; Gemini 3 Pro scored 90.33 on v1.5 in the same paper | Not published |
| Known failures | Cost per page varies with page images; no word coordinates | Text repetition and looping (issue #55), hallucinated text (issue #58), bold and italic formatting mostly lost, runs out of memory on 12 GB GPUs for version 2 |
Sources
- OpenAI file inputs docs
- PaddleOCR-VL-1.5 paper (GPT-5.2 eval)
- HN Mistral OCR thread
- Hugging Face model card
- MarkTechPost on Unlimited OCR 2
- Issue #55 repetition
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 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": "openai/gpt","fallbacks": ["baidu/unlimited-ocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
GPT alternativesUnlimited OCR alternativesBest OCR models in 2026