OCRRouter

GPT vs Typhoon OCR

Pick GPT when you need handwritten notes, custom JSON extraction or document question answering through a hosted API. Pick Typhoon OCR for Thai and Thai-English documents such as government forms and financial reports, where you can self-host an Apache-2.0 model at 2B parameters. The main split is scope: GPT is a general hosted model with variable per-page cost, and Typhoon OCR is a narrow Thai and English OCR model you can run yourself.

GPTopenai/gptTyphoon OCRtyphoon/typhoon-ocr
TypeVision modelOpen source
Price11 credits / page2 credits / page
Per 1,000 pages$11.00$2.00
Typical speed2.4 s / page1.8 s / page
Best forMessy scans, custom instructionsThai documents
InputsPDF, imagesPDF, images
Languages50+ languagesThai, English
Max pages per request100500
Tables✓ Yes✓ Yes
Handwriting✓ YesNo
Math✓ YesNo
Bounding boxesNoNo
ConfidenceNoNo
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 GPT if

  • Choose GPT for handwritten notes: an HN user reported about 5% errors on messy cursive versus about 30% for Mistral OCR.
  • Choose GPT when you need extraction into custom JSON, or question answering over documents, since you can prompt it to do the task directly.
  • Choose GPT if you want a hosted API that takes PDFs up to 50 MB per request and sends both extracted text and page images to the model.

Choose Typhoon OCR if

  • Choose Typhoon OCR for Thai government forms: V1.5 beat GPT-5 and Gemini 2.5 Pro there with a BLEU of 0.870.
  • Choose Typhoon OCR if you need to self-host: weights are free under Apache-2.0 (subject to OpenTyphoon Terms and Conditions) and you pay only for compute.
  • Choose Typhoon OCR for tables and equations, since it outputs tables as HTML with merged cells, equations as LaTeX and marks checkboxes.

Key differences

GPTTyphoon OCR
How you run itHosted API onlySelf-hosted weights, or the Typhoon API as typhoon-ocr-v2
LicenceProprietaryApache-2.0, subject to OpenTyphoon Terms and Conditions
PriceToken-based per model; page images add input tokens, so cost per page is not fixedFree when self-hosted (you pay for compute); API pricing not published on the model card
Output formatsMarkdown, JSON, plain textMarkdown, HTML, LaTeX
OmniDocBenchGPT-5.2: 85.50 on v1.5 and 86.59 on v1.6 (independent)V2 2B: 90.97 on v1.6 (reported by the authors)
LanguagesNot publishedThai and English only evaluated
Known failureNo bounding boxes or confidence scores in the standard responseV2 can loop until the token limit on dense multi-column pages; weakest at key-information extraction (21.0 on ThaiOCRBench)

Sources

What each one is

GPT

General vision model. Good at messy scans and following extraction instructions.

GPT details

Typhoon OCR

Open vision model tuned for Thai and English documents, with Markdown output.

Typhoon 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": "openai/gpt","fallbacks": ["typhoon/typhoon-ocr"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

GPT alternativesTyphoon OCR alternativesBest OCR models in 2026