'xAI sued the state in December. It argued the law violated its free-speech rights under the U.S. Constitution and would force the company to reveal trade secrets about how its AI models are trained'

The law is not about training but what data is used. When you think about it, every other sector has to do this. E.g. the food industry has to declare exactly what they put in their sussages.

Apples to oranges? I dont know

Lets make it 35. We can not trust people who have not raised a kid themselves.

Shit, what age am I gonna have that government-issued kid given to raise?

Are people really happy with zed or is this the vocal minority we see on HN?

I have tried zed multiple times and always uninstalled it after a few days because of the ugly font rendering, slow response time (!), high memory usage (!!) and being very buggy in general. And don't even get me started on their AI-first shenanigans.

This is on Linux, maybe it works better on other platforms.

I'm very happy with zed if I have to compare it to any intellij product

If you're using Windows (+ WSL) then yes, there's a hunch of memory issues currently. Apart from Windows I've never issues though, especially seeing as how intellij indexing takes >1 min and hogs my memory during that period.

The only reason I'm not switching is because Zed's vim bindings are aaalmost perfect. I did not like Vim Mode in VSC or intellij IDE's since it's another keymap I have to learn next to my nvim setup

No, I had issues on native Linux. At one point it used 4x available memory and brought down the entire system.

Maybe this is a vim thing? What do emacs and vscode users feel about zed?

I initially loved Zed because it was so much snappier than VSCode/Cursor, but running several Zed instances made my Ryzen/32gb machine unusable together with Claude because Zed seems such a memory hog. Not using it currently anymore. (Win11)

I tried Zed. It's alright. Unfortunately the devs think it's ok to download and run extra software without asking me (specifically node.js but it would be bad no matter what the software was). I am not willing to use software where the developers have such a cavalier approach to the user's PC, so I went back to Sublime.

Same experience, I heard about them in some podcasts, it all sounded good, performance, from the ground up, rust, collaboration, whatever.

Tried it, it was the worse experience I had with an editor since I started my career… then I tried it again because maybe they figured it out… it’s still bad.

I have done the same multiple times. But for the last 3 months or so. I haven’t touched anything else (well nvim for quick code browsing, but not as editor). I think that might be around the time they fixed the font rendering on Linux.

It’s good. Launches from anywhere from cli with just ‘zed .’ (No logs and artifacts that hold up the terminal). Multiple instances without worry (unlike PyCharm I used before).

It’s good to the point I keep checking to see if there is a “pro” plan, but there isn’t. They just have AI subscriptions.

I've been using it daily for about a year now. I love it. Using most other IDEs feels like a major regression. Multi-buffers are beautifully implemented. I typically use Claude Code if I use an LLM, but their Claude integration is excellent.

Debugging is good. I'd love to see integrated test runners.

I also wish the collaboration was more relevant and fleshed out, but no one I know uses it anyway.

Went to Zed from Sublime and never looked back (I'm on Mac). Never touched VSCode apart of small tests to see that the project setting (format on save, etc) will work for my colleagues. Can't stand it.

I have opened at all times at least 3-4 medium sized front-end projects and at least one large python project and never had memory issues.

I've been using it as a replacement for VSCode/Cursor for the past few months and really like it.

There have been recent changes to the font rendering, it now looks good on my 1440p monitor - when I first tried, it was blurry and instantly gave me a headache. It does seem strange to me that *font rendering* of all things is what a text editor has trouble with. Precise font rendering should be table stakes here.

As for performance/memory usage, it's been top-notch for me on both Linux and macOS.

I had originally switched to try and get away from Microsoft's "only focus on AI features" trend in VSCode, but it seems like Zed has fallen into the same trap.

I agree there are really noticeable issues on Linux. For example, closing file windows is shockingly slow. I'm talking like 500ms+ to close on a high end Ubuntu machine.

The macOS version is where they've invested the most optimizations for sure. None of the issues you've described occur on there. Fwiw, I've been daily driving the editor on my mac for about a year with almost no complaints.

I’m on a mac - I’ve been poking at zed a bit and it’s… fine? I was a die hard Textmate user, but that’s not just dead but decayed at this point, and I’ve yet to find anything as good.

I haven't had this experience (on macOS, it is quite fast), but I have read this exact same complaint about rendering/slowness from other Linux users a handful of times. So I think those might be more platform-specific issues. I do see high memory usage, though. So I enjoy using Zed when I want to use AI, and prefer Sublime when I want to fly solo.

if you would say you are writing about zed, I'd think you were writing about vscode.

I can't believe the latency of browser based editor is acceptable to people

*wouldn't have said

To me it's just a Sublime with AI. Except Sublime doesn't eat my RAM.

Very fast but not enough features to use on a daily basis.

And they lack some extensions I love in VSCode such as Markdown All In One and GitLens. They have a big battle against the large ecosystem of VSCode extensions.

And then there's JetBrains which is on another level of refactoring/features.

I've been trying it out on MacOS for the past couple of weeks and I'm happy with it. I have a fairly vanilla use of my code editors and I find Zed snappier than VSCode. I have not experienced the issues you describe.

I find it okay on macOS. Font rendering isn't the best but palatable.

It used to be the best. Straight to the point, simple, great settings.

Same. MacOS. Tried it. Was okay-ish for several days. But eventually I realised it's worse than WebStorm in basically every aspect, subjectively speaking.

It is slow, it's missing features and it is buggy.

Because it's the only good free IDE for rust.

the only time it's using high memory usage if the project runs like 3-4 LSP's. then it's pretty slow, yes.

Personally I love it. It just feels fast and minimal. I'm on Mac.

Is everyone a content creator these days?

Besides, most of what you mentioned doesn't run on NPU anyway. They are usually standard GPU workload.

The issue is that the consumer strongly associates "AI" with LLMs specifically. The fact that machine learning is used to blur your background in a video call, for example, is irrelevant to the consumer and isn't thought of as AI.

    'People said about the gemini team that moved to xai that they were "amateurs". And yet they delivered in about 1 year with grok4, was SotA for a few weeks'

Wait a minute, this is the same company that is sueing OpenAI for... pretty much this?

Love how they are closing with the CRA reference

CSP=Content Security Policy

Which probably doesn't help much you if you dont know about XSS.

How would this work on other agents? Gemini for example eats tokens like crazy.

Z80 is CISC. This looks like a MIPS.

Funny enough, there is a 32-bit version of Z80 called Z380.

I think you are into something here.

I tried creating an emulator for CPU that is very well known but lacks working open source emulators.

Claude, Codex and Gemini were very good at starting something that looked great but all failed to reach a working product. They all ended up in a loop where fixing one issues caused something else to break and could never get out of it.

When they get stuck, I find adding debug that the model can access helps. + Sometimes you need to add something into the prompt to tell it to avoid some approach at a point.

I’ve been trying to do the same thing as a hobby project to just imagine some “what ifs” with some slight changes to the original 8086 and the 80286.

It just never produces an actually working result without a lot of intervention on my part. (My change was merely changing the paragraph size from 4 bits to 8 bits on the 8086.)

Interesting. When I had Claude write a language transpiler it always checked that tests passed before declaring a feature ready for PR. There was never a case where it gave up on achieving that goal.

Please tell me what CPU it is. I would give it a try. I doubt strongly a very well documented CPU can't be emulated by writing the code with modern AIs.