The LineShine supercomputer at the National Supercomputing Center in Shenzhen. | Photo: Liang Xu/Xinhua via Getty Images Despite trade restrictions, China has reclaimed the title of the world's fastest supercomputer for the first time since 2018. LineShine has pushed El Capitan out of number one on the TOP500 ranking. That's despite strict limits on what high-powered computing components can be sold to China by US firms, which dominate the list, with America holding three of the top five spots. LineShine doesn't even use any GPUs, which are typically the backbone of modern supercomputers.While reaching the peak of the Top500 carries obvious bragging rights, it also serves as a message from the Chinese government to the US. The Trump administration has sought …Read the full story at The Verge.

Local AI used to ask one big question of you: what can your hardware actually take? The answer determined which models you could run, the context length you could push, and whether vision or audio were even on the table. The question still remains, but it's not as black and white anymore because smaller models keep showing up that don't really care about your specs as much.

Using an aging PC build can feel frustrating. It used to boot silently and handle 4K gaming or heavy code compilation without breaking a sweat. Now the fans sound like a jet engine the second you open a browser tab, and games that used to hit a locked 90FPS are suddenly hitching and dipping into the forties.

China's Zhipu AI (Z.ai) released its open-weight GLM-5.2, and some researchers have claimed that it matches Mythos in certain bug-finding and cybersecurity scenarios. While GLM lags behind models from Anthropic and OpenAI in other, more general tasks, it seems that China has dramatically reduced the gap in the capabilities between its models and those of the US.This level of advancement is particularly concerning to the US government, which has worked to restrict China's access to powerful models like Anthropic's Mythos and Fable, as well as the hardware necessary to train and run them. The Trump administration views Mythos and other advan …Read the full story at The Verge.

There are countless reasons why AI-assisted coding tools like Claude Code, Codex, Cursor, Windsurf, and so on, are incredible. They've practically removed the barrier there once was to begin coding, and turned what used to be hours of grinding through documentation into a conversation. You can now go from a chaotic idea spiraling in your head into a fully working prototype within seconds, and a non-technical person no longer needs to sign up to a months-long course just to build the thing they have in mind.

I've been heavily exploring the world of agentic coding over the last few months, and one thing stuck out to me. No matter which harness or tool I used, every session went the same way. I'd define a goal in a fresh chat, lay down the ground rules for bug-fix steps, add any other workflow options I can remember, and remind the agent about the test-then-commit ritual I always follow (but sometimes forget).

For the majority of gamers, diminishing returns become severe. Moving to 8,000 Hz doesn't give you a competitive edge unless you're a professional player or gaming at an extremely high level. Instead, it just introduces a CPU processing tax for the majority as an expensive solution to a problem that the rest of the input pipeline usually dominates.

Pretty much everything costs more these days, from gas to groceries to online services. For instance, the cost of a standard individual YouTube Premium plan was $12 per month when it was first launched in 2018 and remained at that price through most of 2020 — it's now more than $16 in the U.S. Instead of playing along with Google, I decided to proudly cancel my YouTube Premium account and look at other solutions to get my videos on my television.

I'd been an X11 user for decades, so when I saw that the developers of KDE were dropping support in favor of Wayland, I didn't know how to feel. If you were an early Linux user, you couldn't get away from the X Window System unless you wanted to use the command line without a GUI, and we all grew up knowing how to use and tweak it.

Suno has ambitions to be more than just a toy to churn out AI slop, it also wants to be a streaming destination and to break new artists. Spark is their new incubator program for independent artists that provides grants, mentorship, and marketing support.To apply, artists need to be an unsigned singer, songwriter, or producer releasing music under their own name. They also need to agree to some terms and conditions that have raised some eyebrows over on the Suno subreddit. For one, you need to agree to make your songs available on Suno for remixing. That's not necessarily super concerning, but the broad license it grants Suno to your works …Read the full story at The Verge.

A Jellyfin server makes it much easier to consume your media on all the devices you own. By hosting it on a central device on your network, you can stream to any device in your home (or outside of it, with the right setup) without being restricted by the corporations that run modern streaming services.

When I replaced my Ryzen 9 5900X with a 5800X3D a couple of years ago, the performance improvement was obvious almost immediately. Some games, like Valorant and Battlefield 2042, were clearly CPU-limited at 1440p when I had the 5900X, often leaving my RTX 4090 sitting at around 70–80% usage. After the upgrade, both ran significantly better, with GPU usage improving across the board. It was so good that I never felt the need to ditch AM4 unless I wanted the RTX 5090.