OCR models

Turn scanned documents, PDFs and photos into Markdown, HTML tables and LaTeX. These models take a page or element image as input. Available via POST /chat/completions.

OCR guide

Plans

All models

Recommended

• 

  LightOnOCR-3-4B

  LightOnOCR-3-4B

  Copied: LightOnOCR-3-4B

  Context

  32K

  Hosting

  T-Cloud Germany (EU)

• 

  PaddleOCR-VL-1.6

  PaddleOCR-VL-1.6

  Copied: PaddleOCR-VL-1.6

  Context

  128K

  Hosting

  T-Cloud Germany (EU)

Both run on T-Cloud, in the Essential, Professional and Agentic plans. Context windows are 32K and 128K.

OCR

ProviderAll

LightOnPaddlePaddle

CloudAll

T-Cloud

PlansAll

EssentialProfessionalAgentic

All filtersFilters

All filters

Provider

LightOnPaddlePaddle

Cloud

T-Cloud

Input

Image

Output

Text

Streaming

✓ Supported✗ Not supported

Tools

✓ Supported✗ Not supported

Plans

EssentialProfessionalAgentic

Clear allShow 2 models

Showing 2 of 2 models

Sort by Recommended orderModelProviderCloudLocationInput typesStreamingToolsContext (tokens)ReasoningReasoning tokensInput (per 1M tokens)Output (per 1M tokens)Cached input (per 1M tokens)Plans

Model	Cloud	Location	Input types	Output types	Streaming Token-by-token output via Server-Sent Events. Suitable for low-latency, real-time UI.	Tools Function / tool calling (OpenAI-compatible). The model can return structured tool invocations.	Context tokens	Input per 1M tokens	Output per 1M tokens	Cached input per 1M tokens Price per million prompt tokens served from the prefix cache: 15% of the normal input price. T-Cloud-hosted language models only.	Plans	Details
LightOn LightOnOCR-3-4B	T-Cloud Hosted on T-Cloud Public — Telekom's sovereign infrastructure in Germany, within the EU.	Germany (EU)	Image	Text	✓	✗	32K	€0.20	€0.60	€0.03	EssentialProfessionalAgentic	Show details: LightOnOCR-3-4B
PaddlePaddle PaddleOCR-VL-1.6	T-Cloud Hosted on T-Cloud Public — Telekom's sovereign infrastructure in Germany, within the EU.	Germany (EU)	Image	Text	✓	✗	128K	€0.10	€0.40	€0.015	EssentialProfessionalAgentic	Show details: PaddleOCR-VL-1.6