nemotron-ocr-v2

Description

nemotron-ocr-v2 is a state-of-the-art multilingual text recognition model designed for robust end-to-end optical character recognition (OCR) on complex real-world images. It integrates three core neural network modules: a detector for text region localization, a recognizer for transcription of detected regions, and a relational model for layout and structure analysis.

This model is optimized for a wide variety of OCR tasks, including multi-line, multi-block, and natural scene text, and supports advanced reading order analysis via its relational model component. nemotron-ocr-v2 supports multiple languages and is production-ready with a focus on speed and accuracy on both document and natural scene images.

nemotron-ocr-v2 is part of the NVIDIA NeMo Retriever collection, which provides state-of-the-art, commercially-ready models and microservices optimized for the lowest latency and highest throughput.

This model is ready for commercial use.

License and Terms of Use:

GOVERNING TERMS: The trial service is governed by the NVIDIA API Trial Terms of Service. Use of this model is governed by the NVIDIA Open Model License Agreement.

You are responsible for ensuring that your use of NVIDIA provided models complies with all applicable laws.

Model Developer: NVIDIA

Deployment Geography:

Global

Use Case:

This model is designed for high-accuracy and high-speed extraction of textual information from images across multiple languages, making it ideal for powering multimodal retrieval systems, RAG pipelines, and agentic applications that require seamless integration of visual and language understanding.

Release Date:

Build.NVIDIA.com 06/12/2026 via nemotron-ocr-v2
Hugging Face 04/15/2026 via nvidia/nemotron-ocr-v2
NGC 06/12/2026 via nemotron-ocr-v2

Reference(s):

References:

• NVIDIA NIM Documentation

Model Architecture:

Architecture Type: Hybrid detector-recognizer with document-level relational modeling

nemotron-ocr-v2 is available in two variants:

• v2_english — Optimized for English-language OCR with a compact recognizer for lower latency.
• v2_multilingual — Supports English, Chinese (Simplified and Traditional), Japanese, Korean, and Russian with a larger recognizer to accommodate the expanded character set.

Both variants share the same three-component architecture:

• Text Detector: Utilizes a RegNetX-8GF convolutional backbone for high-accuracy localization of text regions within images.
• Text Recognizer: Employs a pre-norm Transformer-based sequence recognizer to transcribe text from detected regions, supporting variable word and line lengths.
• Relational Model: Applies a multi-layer global relational module to predict logical groupings, reading order, and layout relationships across detected text elements.

All components are trained jointly in an end-to-end fashion, providing robust, scalable, and production-ready OCR for diverse document and scene images.

Network Architecture: RegNetX-8GF

Recognizer Comparison

Parameter Counts — v2_english:

Parameter Counts — v2_multilingual:

Input:

Output:

Sample output

ocr_boxes = [[[15.552736282348633, 43.141815185546875],
  [150.00149536132812, 43.141815185546875],
  [150.00149536132812, 56.845645904541016],
  [15.552736282348633, 56.845645904541016]],
 [[298.3145751953125, 44.43315124511719],
  [356.93585205078125, 44.43315124511719],
  [356.93585205078125, 57.34814453125],
  [298.3145751953125, 57.34814453125]]]

ocr_txts = ['The previous notice was dated', '22 April 2016']
ocr_confs = [0.97730815, 0.98834222]

Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA's hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.

Software Integration:

Runtime Engines: TensorRT, PyTorch
Supported Hardware Microarchitecture Compatibility:
NVIDIA Ampere
NVIDIA Blackwell
NVIDIA Hopper
NVIDIA Lovelace
Operating Systems: Linux

The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment. Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment.

Model Version(s)

nemotron-ocr-v2 (variants: v2_english, v2_multilingual)

Short Name: nemotron-ocr-v2

Training and Evaluation Datasets:

Training Dataset

Data Modality: Image
Training Data Collection: Hybrid (Automated, Human, Synthetic)
Training Labeling: Hybrid (Automated, Human, Synthetic)
Training Properties: Trained on a large-scale, curated mix of public and proprietary OCR datasets, focusing on high diversity of document layouts and natural scene images. The training set includes synthetic and real images with varied noise and backgrounds, filtered for commercial use eligibility. Includes scanned documents, natural scene images, receipts, and business documents.

Evaluation Dataset

Evaluation Data Collection: Hybrid (Automated, Human, Synthetic)
Evaluation Labeling: Hybrid (Automated, Human, Synthetic)
Evaluation Properties: Evaluated on OmniDocBench (crop-level) and SynthDoG (page-level, 7 languages) benchmarks.

Evaluation Results

Inference

Acceleration Engine: TensorRT
Test Software: TensorRT
Test Hardware: NVIDIA L40S

Ethical Considerations

NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their supporting model team to ensure this model meets requirements for the relevant industry and use case, and address unforeseen product misuse.

For more detailed information on ethical considerations for this model, please see the Model Card++ Bias, Explainability, Safety & Security, and Privacy Subcards.

Please report model quality, risk, security vulnerabilities or NVIDIA AI Concerns here.