OpenAI

OpenAI vision models for OCR

openai.com

Engines

6

Available

6

Avg Accuracy

37%

Lowest reference cost

~412 pages / $1~$0.00243 per standard page

Available Models (6)

GPT 4.1 Mini

Available

GPT-4.1 Mini — Cost-effective OCR with GPT-4.1's improved accuracy.

Accuracy: 46.2%

Speed: medium

Price: ~412 pages / $1~$0.00243 per standard page

Latency: 2.8s

printed-texthandwriting

multi-language

layout-analysis

GPT 4o Mini

Available

GPT-4o Mini — Affordable vision model. Good for standard OCR tasks.

Accuracy: 32.3%

Speed: medium

Price: ~401 pages / $1~$0.00249 per standard page

Latency: 1.2s

printed-texthandwritingmulti-languagelayout-analysis

GPT 4.1

Available

GPT-4.1 — OpenAI's latest model with improved vision and document understanding.

Accuracy: 36.1%

Speed: medium

Price: ~174 pages / $1~$0.00575 per standard page

Latency: 2.0s

printed-texthandwritingmulti-languagelayout-analysis

O4 Mini

Available

o4-mini — OpenAI reasoning model. Deliberate analysis for complex documents.

Accuracy: 41.7%

Speed: slow

Price: ~72.1 pages / $1~$0.0139 per standard page

Latency: 2.5s

printed-texthandwritingmulti-languagelayout-analysis

o3

Available

o3 — OCR-capable vision model.

Accuracy: 30.4%

Speed: slow

Price: ~54.7 pages / $1~$0.0183 per standard page

Latency: 2.3s

printed-texthandwritingmulti-languagelayout-analysis

o1

Available

o1 — OCR-capable vision model.

Accuracy: No hard score

Speed: slow

Price: ~10.8 pages / $1~$0.0922 per standard page

Latency: 5.2s

printed-texthandwritingmulti-languagelayout-analysis