Skip to content
  • AI at your fingertips
    AI at your fingertips
    AI at your fingertips
    An artistic representation of the Samsung Exynos chip emerging from a smartphone, surrounded by sparkles and bold "AI" lettering.
On-device AI reimagines what is possible with mobile technology.
Generative AI is here to stay, and now you have the capability to use it on your mobile device.
Enjoy an enhanced mobile experience anytime, anywhere.
On-device AI reimagines what is possible with mobile technology.
Generative AI is here to stay, and now you have the capability to use it on your mobile device.
Enjoy an enhanced mobile experience anytime, anywhere.
On-device AI reimagines what is possible with mobile technology.
Generative AI is here to stay, and now you have the capability to use it on your mobile device.
Enjoy an enhanced mobile experience anytime, anywhere.

What is On-device AI?
What is On-device AI?
What is On-device AI?

On-device is what it sounds like: the capability to run AI features directly on mobile devices, without having to rely on the cloud. Until now, cloud-based AI has been the norm, but it requires sending data to the cloud for processing and retrieval, which leads to latency. Security is also a concern since personal data can be transmitted to external servers when using cloud-based AI. However, with on-device AI, these concerns are no longer an issue. Running AI functions directly on the device reduces latency while enhancing privacy and flexibility, all without the need for an internet connection. And the key to it all is powerful neural processing unit (NPU) performance.
On-device is what it sounds like: the capability to run AI features directly on mobile devices, without having to rely on the cloud. Until now, cloud-based AI has been the norm, but it requires sending data to the cloud for processing and retrieval, which leads to latency. Security is also a concern since personal data can be transmitted to external servers when using cloud-based AI. However, with on-device AI, these concerns are no longer an issue. Running AI functions directly on the device reduces latency while enhancing privacy and flexibility, all without the need for an internet connection. And the key to it all is powerful neural processing unit (NPU) performance.
On-device is what it sounds like: the capability to run AI features directly on mobile devices, without having to rely on the cloud. Until now, cloud-based AI has been the norm, but it requires sending data to the cloud for processing and retrieval, which leads to latency. Security is also a concern since personal data can be transmitted to external servers when using cloud-based AI. However, with on-device AI, these concerns are no longer an issue. Running AI functions directly on the device reduces latency while enhancing privacy and flexibility, all without the need for an internet connection. And the key to it all is powerful neural processing unit (NPU) performance.
Image comparing ai using cloud servers with on-device ai
Tap into the potential of AI Tap into the potential of AI Tap into the potential of AI

Harness cutting-edge AI through a powerful NPU. On-device AI has evolved from its initial focus of enhancing photo and video quality, as well as making editing more convenient. Powered by a high-performance NPU that quickly processes complex computations, it converts real-time translations into the user’s voice and generates videos based on their prompts.

Harness cutting-edge AI through a powerful NPU. On-device AI has evolved from its initial focus of enhancing photo and video quality, as well as making editing more convenient. Powered by a high-performance NPU that quickly processes complex computations, it converts real-time translations into the user’s voice and generates videos based on their prompts.

Harness cutting-edge AI through a powerful NPU. On-device AI has evolved from its initial focus of enhancing photo and video quality, as well as making editing more convenient. Powered by a high-performance NPU that quickly processes complex computations, it converts real-time translations into the user’s voice and generates videos based on their prompts.

A person holding a smartphone with a voice assistant interface active on the screen, standing outdoors on a sunlit pathway.

Communicate with the world in your own voice anytime, anywhere. When you start a sentence that requires translation, it instantly translates the sentence and naturally converts it into your voice. Because of this, communication is much more seamless and convenient.

Communicate with the world in your own voice anytime, anywhere. When you start a sentence that requires translation, it instantly translates the sentence and naturally converts it into your voice. Because of this, communication is much more seamless and convenient.

Communicate with the world in your own voice anytime, anywhere. When you start a sentence that requires translation, it instantly translates the sentence and naturally converts it into your voice. Because of this, communication is much more seamless and convenient.

Capture the moment without limits. Simply describe a scene you want to photograph with either text or voice, and your smartphone will automatically create a photo of that scene in that instant. This feature goes beyond photography, too — extending to video recording, slow motion, and other advanced actions.

Capture the moment without limits. Simply describe a scene you want to photograph with either text or voice, and your smartphone will automatically create a photo of that scene in that instant. This feature goes beyond photography, too — extending to video recording, slow motion, and other advanced actions.

Capture the moment without limits. Simply describe a scene you want to photograph with either text or voice, and your smartphone will automatically create a photo of that scene in that instant. This feature goes beyond photography, too — extending to video recording, slow motion, and other advanced actions.

Bring the scenes in your mind to life. By inputting a prompt describing the scene you want to generate, you can easily create videos, based on our powerful NPU, whenever you want. The flexibility and unbelievable capabilities redefine what it means to bring ideas to life.

Bring the scenes in your mind to life. By inputting a prompt describing the scene you want to generate, you can easily create videos, based on our powerful NPU, whenever you want. The flexibility and unbelievable capabilities redefine what it means to bring ideas to life.

Bring the scenes in your mind to life. By inputting a prompt describing the scene you want to generate, you can easily create videos, based on our powerful NPU, whenever you want. The flexibility and unbelievable capabilities redefine what it means to bring ideas to life.

Unfold your 2D world in 3D. To visualize how the item in front of you will fit into your desired surroundings, simply take a photo and transform it into 3D. You can also set a background for the generated 3D image and see how well it pairs with the scene. The high performance NPU drives this transformation, providing a seamless 3D experience. Additionally, the AMD Architecture-based Xclipse GPU leverages ray tracing to accurately reproduce light paths, including reflections, refractions, and shadows, for more realistic simulations.

Unfold your 2D world in 3D. To visualize how the item in front of you will fit into your desired surroundings, simply take a photo and transform it into 3D. You can also set a background for the generated 3D image and see how well it pairs with the scene. The high performance NPU drives this transformation, providing a seamless 3D experience. Additionally, the AMD Architecture-based Xclipse GPU leverages ray tracing to accurately reproduce light paths, including reflections, refractions, and shadows, for more realistic simulations.

Unfold your 2D world in 3D. To visualize how the item in front of you will fit into your desired surroundings, simply take a photo and transform it into 3D. You can also set a background for the generated 3D image and see how well it pairs with the scene. The high performance NPU drives this transformation, providing a seamless 3D experience. Additionally, the AMD Architecture-based Xclipse GPU leverages ray tracing to accurately reproduce light paths, including reflections, refractions, and shadows, for more realistic simulations.

a user interface for an AI image generation tool with a completed prompt for creating an image of a whale flying above the clouds
Text-to-image generation Text-to-image generation Text-to-image generation

Generate images from text, in real time. You can input a prompt for the image you would like to be generated, and the device will rapidly produce it.

Generate images from text, in real time. You can input a prompt for the image you would like to be generated, and the device will rapidly produce it.

Generate images from text, in real time. You can input a prompt for the image you would like to be generated, and the device will rapidly produce it.

a before-and-after scenario where an AI tool has altered the color of a hat and expanded the background of a photo set
In-painting & Out-painting In-painting & Out-painting In-painting & Out-painting

Apply modifications that look natural. With “in-painting” and “out-painting”2, you can change a dog’s face in an image to a cat or expand an image to the area not visible in the frame, and it will look as natural as if you had taken the picture that way.

Apply modifications that look natural. With “in-painting” and “out-painting”2, you can change a dog’s face in an image to a cat or expand an image to the area not visible in the frame, and it will look as natural as if you had taken the picture that way.

Apply modifications that look natural. With “in-painting” and “out-painting”2, you can change a dog’s face in an image to a cat or expand an image to the area not visible in the frame, and it will look as natural as if you had taken the picture that way.

Message suggesting dinner with Shakespearean-style AI reply options.
Tone & Grammar fix Tone & Grammar fix Tone & Grammar fix

Edit phrases and change tone with ease on your device. As an example, Google’s Magic Compose analyzes information and then allows you to simply correct text or modify tone of voice accordingly. This means you can craft messages with whatever tone you like – whether it be excited, formal or poetic.

Edit phrases and change tone with ease on your device. As an example, Google’s Magic Compose analyzes information and then allows you to simply correct text or modify tone of voice accordingly. This means you can craft messages with whatever tone you like – whether it be excited, formal or poetic.

Edit phrases and change tone with ease on your device. As an example, Google’s Magic Compose analyzes information and then allows you to simply correct text or modify tone of voice accordingly. This means you can craft messages with whatever tone you like – whether it be excited, formal or poetic.

Fine-tune your AI Fine-tune your AI Fine-tune your AI

Exynos AI Studio is a toolchain that generates AI models developed by customers into a form that can run on Exynos mobile processors. It provides lightweight optimization technology tailored for Exynos mobile processors, optimizing both the structure and parameters of AI models while efficiently balancing computational and memory resources — all while maintaining high inference accuracy.

It supports a wide range of advanced models, including not only multimedia models for image, voice, and video processing but also large vision models (LVMs), large language models (LLMs), and large multimodal models (LMMs), covering an extensive range of generative models.

Exynos AI Studio is a toolchain that generates AI models developed by customers into a form that can run on Exynos mobile processors. It provides lightweight optimization technology tailored for Exynos mobile processors, optimizing both the structure and parameters of AI models while efficiently balancing computational and memory resources — all while maintaining high inference accuracy.

It supports a wide range of advanced models, including not only multimedia models for image, voice, and video processing but also large vision models (LVMs), large language models (LLMs), and large multimodal models (LMMs), covering an extensive range of generative models.

Exynos AI Studio is a toolchain that generates AI models developed by customers into a form that can run on Exynos mobile processors. It provides lightweight optimization technology tailored for Exynos mobile processors, optimizing both the structure and parameters of AI models while efficiently balancing computational and memory resources — all while maintaining high inference accuracy.

It supports a wide range of advanced models, including not only multimedia models for image, voice, and video processing but also large vision models (LVMs), large language models (LLMs), and large multimodal models (LMMs), covering an extensive range of generative models.

A glowing 3D cube with blue neural network-like dots on a dark background, Exynos AI Studio.

Pioneering the future of mobile AI
Pioneering the future of mobile AI
Pioneering the future of mobile AI

Samsung has continuously enhanced the mobile NPU performance of Exynos, paving the way for the on-device AI era. This mobile NPU technology has now evolved to the point where it can enable generative AI directly on the device for not only images and language, but also audio and video. By supporting the optimization of various AI models for Exynos through Exynos AI Studio, Samsung is providing users with a highly sophisticated on-device AI experience across both hardware and software. Continuing this progress, the Exynos team at Samsung has been and will continue to work closely with smartphone manufacturers and other partners in the ecosystem to develop various use cases for on-device AI. For example, our team is collaborating with Google on Android AICore and Gemini Nano to enable the implementation of generative AI, such as text generation, directly on the device.
Samsung has continuously enhanced the mobile NPU performance of Exynos, paving the way for the on-device AI era. This mobile NPU technology has now evolved to the point where it can enable generative AI directly on the device for not only images and language, but also audio and video. By supporting the optimization of various AI models for Exynos through Exynos AI Studio, Samsung is providing users with a highly sophisticated on-device AI experience across both hardware and software. Continuing this progress, the Exynos team at Samsung has been and will continue to work closely with smartphone manufacturers and other partners in the ecosystem to develop various use cases for on-device AI. For example, our team is collaborating with Google on Android AICore and Gemini Nano to enable the implementation of generative AI, such as text generation, directly on the device.
Samsung has continuously enhanced the mobile NPU performance of Exynos, paving the way for the on-device AI era. This mobile NPU technology has now evolved to the point where it can enable generative AI directly on the device for not only images and language, but also audio and video. By supporting the optimization of various AI models for Exynos through Exynos AI Studio, Samsung is providing users with a highly sophisticated on-device AI experience across both hardware and software. Continuing this progress, the Exynos team at Samsung has been and will continue to work closely with smartphone manufacturers and other partners in the ecosystem to develop various use cases for on-device AI. For example, our team is collaborating with Google on Android AICore and Gemini Nano to enable the implementation of generative AI, such as text generation, directly on the device.
An image of Exynos processor over a wide net that symbolizes neural network
  • * All product specifications reflect internal test results and are subject to variations by user’s system configuration. Actual performance may vary depending on use conditions and environment.
  • * All functionality, features, specifications and other product information provided in this document including, but not limited to, the benefits, components, performance, availability, and capabilities of the product are subject to change without notice or obligation.
  • * All images shown are provided for illustrative purposes only and may not be an exact representation of the product or images captured with the product. All images are digitally edited, modified, or enhanced.
  • * Samsung reserves the right to change images and specifications at any time without notice. Measurements are approximate. All data were deemed correct at time of creation. Samsung is not liable for errors or omissions.
  • 1) Vision Language Model
  • 2) In-painting refers to filling in or replacing different parts of an image, while out-painting refers to extending an image past its original size.