As technologies change and adapt, we’re often left with seemingly useless junk that has nowhere to go. Certainly anyone still sitting on a pile of floppy disks feels this way sometimes, but odds are anyone who owns a mining ASIC or an NFT can attest to that as well. The trillions of dollars flowing into GPU-based data centers will likely become the next victim of this trend, so if you want to capitalize on the losses of some venture capitalist you’ll want to figure out a way to get GPUs meant for a server into your desktop doing useful work.

Of course, calling these devices GPUs is a bit of a stretch compared to the Radeon and GeForce cards many of us are used to using for gaming. These don’t even have a PCIe slot or video output, after all. But, as [Oscar] notes, the VRAM and GPU cores are very real and can still do useful work. An adapter board is able to mate a Tesla V100 SXM2 16 GB GPU to a standard PCIe slot, which solves the first problem, but the major downside from there is that the cooling fan for this unit was literally deafeningly loud. At 82 dB it was about as loud as a lawnmower, which is fine in a server rack but not great in a bedroom. [Oscar] found a way to tamp down the fan speed, making it usable in a home.

Without video output, the utility of these cards mainly comes from adding VRAM and compute for tasks that benefit from parallel computing. Using tensor splitting, [Oscar] is running a local LLM with this card alongside his RTX 4080, providing 32 GB of VRAM on his NixOS system. With his benchmarking tests, the LLM sports impressive stats for a self-hosted model, ranking somewhere around Claude Sonnet 4.6. What’s even more impressive is that this is all done for around £200, and with the rate the various LLM companies are ratcheting up pricing could pay itself back very quickly. If trading off performance for cost is acceptable, though, it’s possible to run local models on much less powerful hardware as well.

[Simon Willison] ported the Moebuis 0.2B image inpainting model to run locally in a web browser.  The web tool simply requires a user to provide an image, mark a section of it to be removed, and the model will do it’s best to patch up the missing area. The project was handled by Claude Code as an experiment in how things in the AI coding world have evolved, but more on that in a moment.

The existence of this tool shows that it’s possible for this kind of image editing to be done on the client side, running entirely locally with no reliance on remote services or server-side GPU resources. The online demo (GitHub repository here) is available if you want to try it out, but be warned it triggers a 1.27 gigabyte download of the required model on the first run.

What’s also interesting is [Simon]’s write-up, because he used the project as an opportunity to learn what has changed in the realm of AI coding agents. [Simon] is a software developer but in this project he didn’t personally write any of the code. One may think that means he didn’t learn anything other than how to use the tools, but that’s not quite true.

He learned it’s possible to convert a PyTorch-based model to ONXX, that the converted model can run in supported browsers using local WebGPU acceleration, and that the CacheStorage API will work on large files. Last but not least, he learned Claude Opus 4.8 is capable of handling such a project pretty much autonomously, and even created an informative document explaining the underlying architecture.

One may consider AI coding agents to be disasters waiting to happen, but it’s also true that the landscape is changing quickly, and write-ups like [Simon]’s give a helpful peek at those developments.