Razer Just Made Its Free AI Toolkit Way More Useful for Creators

Razer just dropped a pretty big update to its AI toolkit, and it is actually kind of interesting even if you are not a hardcore developer. You might know Razer for those glowing keyboards and gaming laptops that cost a small fortune. But the company has been quietly working on this free open source thing called AIKit. Back in January at CES, it was mostly for text stuff.

Now? It can handle images, video, and audio. Plus it finally works on Arm64 chips, which is a big deal if you are into newer hardware like those NVIDIA Grace setups.

The whole point of AIKit is letting you run AI models directly on your own computer instead of paying a cloud subscription every single month. No surprise fees. No internet drama. It automatically figures out your GPU settings and does the boring tuning work for you. So whether you are using a beefy desktop or a Razer Blade laptop, the experience stays the same. That is pretty neat for anyone who hates dealing with configuration files.

But here is where it gets fun.

Razer ran this April Fools’ campaign called AVA Mini. Basically you uploaded a photo of your real pet, and the system spat out a custom 3D AI pet companion in a few seconds. Cute, right? Behind the scenes though, that campaign was actually a giant stress test for AIKit. Over five days at the end of March, they generated more than 11,000 images. Usually if you try to do that with normal cloud services like standard APIs, you are paying anywhere from three to fifteen cents per image. That adds up fast when you are doing thousands of them.

So Razer teamed up with this platform called Akash Network.

Think of it as a decentralized marketplace for GPUs. People rent out their own graphics cards. Razer took the exact same AIKit setup they used during testing and just dropped it onto Akash’s network of RTX 4090 and 5090 cards. The result? Each image cost only one cent. That is up to 15 times cheaper than the usual route. The turnaround time was about 3.2 seconds per image, and at peak they were cranking out 30 images every minute. Zero human intervention across the whole five days.

What makes this smart is that you can build something on your local machine and then scale it up using the exact same tools. No rewriting everything. No lock in to one cloud provider. Razer also put AIKit on Akash’s web console now, so you do not even need to mess with command line stuff if you do not want to. Just click around in a browser.

Moving forward, Razer says they are adding voice and video support.

So you will eventually be able to build apps that talk back or generate video clips using the same workflow. For a brand best known for RGB lighting and esports, this is a solid move into actual developer tools. And because it is free and open source, anyone can go grab it on GitHub right now.

Even if you are just messing around for a school project or a side hustle, it is worth a look.