Gemma-4-26B-A4B-IT

Description

gemma-4-26B-A4B-it is a Google-developed multimodal instruction-tuned Mixture-of-Experts model for text, image, and video understanding. The model supports long-context reasoning up to 256K tokens, coding, tool calling, multilingual conversation, multi-turn conversation, visual reasoning, and interleaved multimodal input. The 26B A4B configuration has approximately 25.2B total parameters and approximately 3.8B active parameters per token.

This model is ready for commercial/non-commercial use.

Third-Party Community Consideration:

This model is not owned or developed by NVIDIA. This model has been developed and built to a third-party's requirements for this application and use case; see link to Non-NVIDIA gemma-4-26B-A4B-it Model Card.

License and Terms of Use:

GOVERNING TERMS: The NIM container is governed by the NVIDIA Software License Agreement and the Product-Specific Terms for NVIDIA AI Products. Use of the model is governed by the NVIDIA Open Model Agreement.

Additional Information: Apache 2.0, Gemma Terms of Use, and Gemma Prohibited Use Policy.

Deployment Geography:

Global

Use Case:

Use Case: Developers and enterprises can use this model to build multimodal assistants, coding copilots, document and image understanding workflows, video understanding workflows, long-context analysis systems, tool-calling agents, and multilingual conversational applications.

Release Date:

NGC 06/01/2026 via NGC

HuggingFace: 04/30/2026 via nvidia/Gemma-4-26B-A4B-NVFP4

Reference(s):

References:

• Google gemma-4-26B-A4B-it Model Page
• NVIDIA Gemma-4-26B-A4B-NVFP4 Model Page
• Google Gemma documentation

Model Architecture:

Architecture Type: Transformer Network Architecture: Multimodal Mixture-of-Experts conditional generation architecture using Gemma4ForConditionalGeneration, 30 layers, 8 active experts per token, 128 total experts, one shared expert, sliding-window attention, and a vision encoder with approximately 550M parameters. Total Parameters: 25.2B Active Parameters: 3.8B Vocabulary Size: 262K tokens Base Model: google/gemma-4-26B-A4B

Input:

Input Types: Text, Image, Video Input Formats: Text: String; Image: Red, Green, Blue (RGB); Video: mp4, mov, webm Input Parameters: One Dimensional (1D), Two Dimensional (2D), Three Dimensional (3D) Other Input Properties: Text prompts, chat messages, code, tool-use instructions, image inputs, interleaved image-and-text inputs, and video inputs. The model supports context length up to 256K tokens. Input Context Length (ISL): 256K tokens

Output:

Output Types: Text Output Format: String Output Parameters: One Dimensional (1D) Other Output Properties: Generated natural language, code, structured text, tool-call arguments, reasoning responses, image-grounded answers, and video-grounded answers.

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: vLLM

Supported Hardware:

• NVIDIA B200 Tensor Core GPU
• NVIDIA H20 Tensor Core GPU
• NVIDIA H200 Tensor Core GPU
• NVIDIA L40S GPU
• NVIDIA DGX Spark

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)

nvidia/Gemma-4-26B-A4B-NVFP4

Training, Testing, and Evaluation Datasets

Training Dataset

Data Modality: Text, image, video, and code Text Training Data Size: Undisclosed Image Training Data Size: Undisclosed Video Training Data Size: Undisclosed Data Collection Method by dataset: Undisclosed Labeling Method by dataset: Undisclosed Properties: The training corpus includes large-scale multimodal data spanning web documents, code, mathematics, and images, with multilingual coverage and a January 2025 cutoff.

Testing Dataset

Data Collection Method by dataset: Undisclosed Labeling Method by dataset: Undisclosed Properties: Undisclosed

Evaluation Dataset

Data Collection Method by dataset: [Hybrid: Automated, Human] Labeling Method by dataset: [Hybrid: Automated, Human] Properties: The evaluation benchmarks included text, coding, reasoning, long-context, and vision benchmarks including MMLU Pro, AIME 2026, LiveCodeBench v6, Codeforces ELO, GPQA Diamond, Tau2, HLE, BigBench Extra Hard, MMMLU, MMMU Pro, OmniDocBench, MATH-Vision, MedXPertQA MM, and MRCR v2.

Inference

Engine: vLLM Test Hardware:

• NVIDIA B200 Tensor Core GPU
• NVIDIA H20 Tensor Core GPU
• NVIDIA H200 Tensor Core GPU
• NVIDIA L40S GPU

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. Developers should work with their internal developer team to ensure these software components meet requirements for the relevant industry and use case and address unforeseen product misuse.

Please make sure you have proper rights and permissions for all input image content; if image includes people, personal health information, or intellectual property, the image generated will not blur or maintain proportions of image subjects included.

Users are responsible for model inputs and outputs. Users are responsible for ensuring safe integration of this model, including implementing guardrails as well as other safety mechanisms, prior to deployment.

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

Get Help

Getting started with the NIM

Deploying and integrating the NIM is straightforward thanks to our industry standard APIs. Visit the NIM Model page for release documentation, deployment guides and more.

NVIDIA Developer Community Forum

Get access to community knowledge base articles and support cases (https://forums.developer.nvidia.com/)