Hacker Newsnew | past | comments | ask | show | jobs | submit | Bigsy's commentslogin

Hmm I'm not sure this is really the case I have the pro level of each, and GPT 6.1 Sol feels virtually unlimited to me atm, doesnt feel the same with opus and its probably only a few percentage points better and not in all situations.


No integration to Claude subscriptions seems like it probably makes dead in the water which is a shame


Hello, As my colleagues already replied earlier to someone else here, we tried to make this work, but unfortunately hit a hard stop with Anthropic: their current policies do not allow us to support this integration, and we were not able to negotiate an exception


How come I can do this in Zed?


You can actually plug Zed's ACP into Air's acp.json file to draw subscription tokens:

{ "agent_servers": { "Claude Code": { "command": "bunx", "args": ["@zed-industries/claude-code-acp"], "env": {} } } }

Swap bunx with npx -y if necessary


Brew install is broken

It seems pretty rubbish I have to say, its recommending me loads of qwen 2.5 which are really old and I'm easy running qwen3.5 and 3.6 models on this mac at decent quants


AI slop quality software for ya.

“I release software now, good luck everyone”


For MLX I'd guess.


That also comes upstream from llama.cpp https://github.com/ggml-org/llama.cpp/discussions/4345



Does this have a CLI only interface?


Yes. You could also look at the README.md.


Does it have custom layouts? I can see it mentioned, I have rotated monitors so none of the standard ones really suit.


they are supported under the hood but the UI for creating them isnt built yet but its on the roadmap.I'll prioritize it (:


Seems like the worst of both worlds. I wouldn't swap my mac hardware for anything so have to do my best to live with macOS but I would take any other OS over it. I just try my best to say inside a terminal if I can.


This. I don’t care much for MacOS. It pretty much gives me a terminal and browser and gets out of the way. But, the hardware, and drivers…. I can reliably shut the lid and open up at a different place and continue working. The battery last atleast a full working day. Displays text sharply. Touchpad works like no other.

I can’t imagine dealing with Linux without these conveniences.


I daily-drive an M1 MBA and a ThinkPad Z13 Gen 1 running Linux (Bazzite), and the experience (in terms of convenience and reliability) is identical for the most part. In fact I actually prefer the screen on my ThinkPad, the 1200p OLED display is just so much more crisper and vibrant, it's been great for gaming and media consumption.

In saying that, the touchpad experience on the MBA is a touch better, and of course, the battery life is much better on the MBA (as is the thermal effeciency).

In spite of these minor shortcomings, I'm super happy with my ThinkPad in terms of just how stable and reliable it's been under Bazzite/KDE, like never once have I had any issues with the suspend-resume functionality - something that even Windows machines struggle with every now and then.

If only the Snapdragon ThinkPads had first-class Linux support like the x86 ones do... I reckon they can come pretty close to the MacBooks in terms of battery life, unfortunately they're not quite there yet.


>> Seems like the worst of both worlds

Before the Silicon Mac chips, installing Linux on Mac hardware was the best of both worlds. Asahi doesn't work for me so far, sadly.


Deepmind are working on solving the plasma control issue at the moment, I suspect they're probably using a bit of AI.... and I wouldn't put it past them to crack it.


They did/are(?).

> Accelerating fusion science through learned plasma control

https://deepmind.google/discover/blog/accelerating-fusion-sc...

(2022)


This is the thing with AI: We can always come up with a new architecture with different inputs & outputs to solve lots of problems that couldn't be solved before.

People equating AI with other single-problem-solving technologies are clearly not seeing the bigger picture.


Can we? Why haven't we, then? What are the big problems that were unsolvable before and now we can solve them with AI?

Auto-tagging of photos, generating derivative images and winning at Go, I will give you. There's been some progress on protein folding, I heard?

Where's the 21st century equivalent of the steam locomotive or the sewing machine?


Right or wrong you don't think it was unduly combative right off the bat? Manners cost nothing.


It’s not how I would have responded either, but people are entitled to their own ways of communicating.


Clojure would be awesome


It looks exactly like the battery venting.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: