Amuse is a free AI art collaboration with Tensorstack, which allows you to generate original AI art, editing, and even videos using your own PC, updated with new features in video generation. – A new optimized model of AMD Ryzen processors.
However, what’s important about Amuse is that it also provides access to the generation AI models of not only AMD, but all processors. If you have an Intel Core PC with an NVIDIA discrete GPU, that’s fine. Download the latest Amuse 3.0 version and try out the range of your features.
Amuse debuted in 2024 as a collaboration between AMD and Tensorstack. In a way, Amuse is AMD’s answer to Intel’s AI Playground, combining AI Art Generation with video and integrated LLM chatbots (although video creation didn’t work on AI Playground when I tried it recently). Amuse currently lacks LLM functionality, but it has access to many AI ART models, including models that are explicitly optimized for AMD processors. TensorStack also adds new features such as photo filters, local video creation, AI-powered video generation, and Restyling.
Both Amuse and AI Playground offer alternatives to cloud-based services that allow you to generate AI ART using powerful servers, such as Midjourney, Microsoft Copilot, and Google Gemini. Amuse runs locally on your PC. This means you have the privacy to create what you need. Amuse ships in terms of service that prevents you from creating NSFW art. Many people hate AI art, but you can simply use Amuse to see what artificial intelligence can generate.
Foundry
However, AMD recommends its own hardware, but it does not require an AMD Ryzen processor to run it. Amuse is designed to use standalone AMD discrete GPUs like the Radeon 9700 series, but is designed to use the Ryzen AI 300 notebook, especially the AI Powerhouse The Ryzen Ai Max and its video ram gobs. If you own it, Amuse should be one of your first stops when looking for AI art.
AI Art works best with more VRAM, and within Amuse 3.0 you’ll find several AI generation models designed specifically for AMD’s RDNA and XDNA GPUs and NPUs. Specifically, AMD says that AMD-optimized models can generate images up to 4.3 times faster with the Radeon 9700 XT’s FP16 accuracy, and up to 3.3 times faster with the Ryzen AI processor.
Amuse’s performance is dramatically different
AMD recommends owning a PC with Ryzen AI Max+ 395 (64GB of RAM), and a PC with Ryzen AI 9 HX 370 with 32GB of RAM, RX 7900 XTX, 7900 XT, 7900 GRE, or 7800 XT GPU. These are not strict restrictions. I have tried the build on two Ryzen AI 365 PCs. One has 32GB of RAM and the other has 24GB of RAM. (I previously handed the ASUS Rog Flow Z13 (2025) Game Tablet Review Unit to the video team with a Ryzen AI Max+ processor.)
I ran both, but more RAM translated into better performance. This is because you need to manually adjust the available VRAM via AMD’s adrenaline software or the BIOS on your PC to maximize the available VRAM. (This option is not available for dedicated GPUs.)
Foundry
Measure performance in AI ART applications depends on competent hardware. However, the model itself makes a difference. The less refined model runs faster, but produces lower quality images. With AMD recommendations, we tested how fast the prompts for stable diffusion were using both the standard model and the same prompt as the AMD optimized version. Performance has improved dramatically. I needed Amuse 3.0 and my Ryzen AI 365/32GB laptop 685 seconds (11.42 minutes). Using the Ryzen-Optimized model it took me only 270 seconds, or 4.5 minutes. In other words, the AMD estimate was not too far.
But that’s still quite poor. I just used the default StabledIffusion XL Turbo model optimized for the Ryzen AI NPU and sent out four small cartoon raccoons in 37 seconds, but didn’t look half bad. The model is really important.
It’s easy, but refined
As anyone who has used Photoshop or the Open-Source GNU Image Manipulation program knows, the power of an application is not only that it can be done, but it is easy to use.
Creating AI art doesn’t just come up with the right explanation. A great refined model that can generate realistic images, powerful hardware that can generate art without having to wait too long, and a user interface that can start quickly, but offers more refined adjustments for those who want it. I liked the simplicity of AI Playground. However, Amuse has several options: image generation, image or video filtering, and the ability to “design” art based on drawings. The latter is essentially a Windows-specific Coctor application.
Foundry
Like an AI playground, everything is set to be as simple as possible. If Amuse needs to download anything, simply explain why, tell them how much space they will consume, then ask for permission.
A small button in the lower left corner promotes “Expert Mode” and unlocks more advanced features. Here you can download more AMD-optimized models that AMD is now available. (Affordable labels with the AMD logo.) You can also select multiple models for image or video generation.
Expert models allow you to navigate between these models for image generation. (You’ll need to manually load each one, such as replacing the player’s DVD.) However, it’s not immediately clear which models you’ll need to download for use in video generation, so you can improve your search.
Amuse generates images. Images can be used as save, edit, or upscaling, or as an origin for AI-generated videos. Unlike some tools like Adobe Firefly, Amuse does not label an image as AI-generated.
Foundry
Honestly, I was not impressed with whether many of the AI images the models generated used “standard” stable diffusion models or an optimized version for AI. However, there are plenty of amuse offers to help you tweak the output. It’s worth playing with it to see what works for you.
However, I liked the text-to-video feature in Amuse. At least on the model I chose, the default resolution was only 512 x 512, with enough keyframes to create a video for a few seconds. I didn’t want to destroy the laptop so I chose not to push it to its limits (1024 x 1024 “HD” video and a few minutes long).
However, the images were largely consistent and generally provided what I wanted – unless I pushed for more than 10 seconds. It was clear that the generated art was then made from several clips, but it wasn’t always the same. Some of the competing cloud-based models, such as SORA and Google’s VEO, are more cohesive.
Foundry
Is Amuse okay? Yes, and for everyone
Compared to Amuse, Intel’s AI Playground is a little more simple and a little more limited, but a little more welcome. It’s limited to a subset of Intel processors, but it’s a good tool to see what AI has to offer. The Amuse app is definitely aimed at enthusiasts, but the user interface is not particularly overwhelming. Keep in mind that this app may offer features that go beyond that, such as feature extraction.
Amuse is not perfect. I hope it’s “EZ mode” that offers a little more directionality for more advanced features without having to look for “Expert Mode” at the bottom right. Similarly, I would like to see some more tooltips and support to understand what some of the more advanced features can do. However, Amuse is free, and will be installed quickly and will give you access to specialized models of AMD Ryzen processors.
Better yet, you don’t need to be the owner of Ryzen to use it. This will improve Amuse for a period of time, one of the better suited AI art apps.
