artdigital

I’m equally surprised to see these posts pop up everywhere on X, GitHub and now also HN. Am I that old that SSHing into a server through a VPN is such a novel concept nowadays?

Your point here being? In that case people will complain, and the next release will have it included into the setting

I switched my subscription from Claude to ChatGPT around 5.0 when SOTA was Sonnet 4.5 and found GPT-5-high (and now 5.2-high) so incredibly good, I could never imagine Opus is on its level. I give gpt-5.2-high a spec, it works for 20 minutes and the result is almost perfect and tested. I very rarely have to make changes.

It never duplicates code, implements something again and leaves the old code around, breaks my convention, hallucinates, or tells me it’s done when the code doesn’t even compile, which sonnet 4.5 and Opus 4.1 did all the time

I’m wondering if this had changed with Opus 4.5 since so many people are raving about it now. What’s your experience?

Claude - fast, to the point but maybe only 85% - 90% there and needs closer observation while it works

GPT-x-high (or xhigh) - you tell it what to do, it will work slowly but precise and the solution is exactly what you want. 98% there, needs no supervision

Btw, Zed has a similar setting to turn off all AI features :)

I bought a bigger nicer TV (the biggest I ever bought) about 4 years ago and it still brings me joy to this day every time I use it

So no, not only 6 weeks

Getting only 20 minutes of usage with a $240/mo plan is a bit ridiculous. How much usage did you get on 2.5-pro? Is it comparable to Claude Max or ChatGPT Pro on the CLI? So a weekly limit but in reality very hard to hit and mostly 'unlimited' unless very heavy usage?

Claude doesn’t gaslight me, or flat out refuses to do something I ask it to because it believes it won’t work anyway. Gemini does

Gemini also randomly just reverts everything because of some small mistake it found, makes assumptions without checking if those are true (eg this lib absolutely HAS TO HAVE a login() method. If we get a compile error it’s my env setup fault)

It’s just not a pleasant model to work with

Huh really? It’s the exact opposite of my experience. I find gpt-5-high to be by far the most accurate of the models in following instructions over a longer period of time. Also much less prone to losing focus when context size increases

Are you using the -codex variants or the normal ones?

Sadly Grok also started doing that recently. Previously it was much more to the point but now got extremely wordy. The question in the end is a key giveaway that something under the hood has changed when the version number hasn’t

I’m not paying for the human that made the app. I’m paying for the app, aka for an advertised thing with an advertised feature set for a specific price. If I deem the value I get for the price worth it, I will purchase the thing from you.

I will however not pay you monthly just because “the dev needs to eat too” if there is no service provided that justifies the monthly ongoing cost.

I’m with the parent on this. I don’t mind subscriptions if a service is provided that justifies the recurring cost. If it’s a local offline app then I don’t see it justified. Price it accordingly or at least give an option for one-time.

But yes, sub vs non-sub model is a very divisive topic. Personally would never subscribe to something like a offline local todo list

I do a fair bit of music and have never seen a CLAP plugin in the wild

I just purchased a MacBook Air on the spot to bridge the week and returned it when I got my MBP back

Also both times I did it, it took 2-3 days, not the quoted 1+ weeks

So is this their version of Jules / Codex / Copilot agent? Aka autonomous agent in the cloud you give a task and it spits out a PR a bit later?

It’s interesting how all the LLMs slowly end up with the same feature set and picking one really ends up with personal preference.

Me as a dev am happy that I now have 4 autonomous engineers that I can delegate stuff to depending on task difficulty and rate limits. Even just Copilot + Codex has made me a lot more productive

Also rip to all the startups that tried to provide “Claude in the cloud”, though this was very predictable to happen

I have all my small stuff on a Vultr managed k8s with the cheapest nodes.

Big fan of Vultr, I like them a lot, but got bare metal stuff Hetzner is going to be cheaper

Your link 404s

Been thinking about fully self hosting my small stuff for a bit but a mini managed k8s ($10) is just so convenient. I build a new docker image for some thing, add a manifest file to my k8s repo and apply it - my “cluster” spins up a new pod, I can easily hook it up to tailscale or Cloudflare warp to make it accessible, and it’s available

Been using this approach for the past years and if something gets bigger, I move the container to fly or a different k8s cluster in a couple hours max

On my bigger k8s I can then easily add more nodes or scale up pods depending on need, and scale them back down when idle.

Still the main issue with any setup I see is the database. No matter what I use I’d either have a managed Postgres somewhere, or something like litestream, and if that’s not in the same data center it’s gonna add latency sadly

Yes if it hits 79%, you can swap it under AppleCare+ at the Apple Store. My MBP M1 Max is on the second replacement and still great!

I use Al Dente to further optimize battery and calibrate it now and then

When I handed in my M1 Max MBP for repair, I bought a M3? Or M4 MBA to return when I get my MBP back, or to keep if I really notice a difference.

Used it for a week and came to the same conclusion, I felt absolutely no difference in day to day usage except that the MBA is nice and slim. And better battery.

I fail to see how comparing Jules to Claude Code is relevant. They’re completely different.

A good Jules comparison would be OpenAI Codex.

For a Claude Code Google equivalent there’s Gemini Code Assist CLI

I’d say GPT-5 is the best in following and remembering instructions. After an initial plan it can easily continue with said plan for the next 30-60 minutes without human intervention, and come back with a complete working finished feature/product.

It’s honestly crazy how good it is, coming from Claude. I never thought I could already pass something a design doc and have it one-shot the entire thing with such level of accuracy. Even with Opus, I always need to either steer it, or fix the stuff it forgot by hand / have another phase afterwards to get it from 90% to 100%.

Yes the Codex TUI sucks but the model with high reasoning is an absolute beast, and convinced me to switch from Claude Max to ChatGPT Pro

Gemini is an excellent collaborator?

It’s the one AI that keeps telling me I’m wrong and refuses to do what I ask it to do, then tells me “as we have already established, doing X is pointless. Let’s stop wasting time and continue with the other tasks”

It’s by far the most toxic and gaslighting LLM

Mind explaining?

Try the uncensored/jailbroken variants like openai-gpt-oss-20b-abliterated-uncensored-neo-imatrix

I just tried to ask it how to make crystal meth and it generated a very detailed step by step guide

Codex had MCP support for a long long time

AFAIK Maestral doesn’t have block level diff and sync which is the main reason (for me) to use Dropbox: their superior sync engine

I have Maestral (and rclone) running on my raspbi but wouldn’t use it for things like sparsebundle syncing

Oh what, there’s a Helix mode now?

Been wanting to learn Helix more and using it for small edits but never saw a Helix mode in any editor yet

Cloudflare WARP (1.1.1.1 tunnel or Zero Trust) run by default on MASQUE

Then don’t be a hoarder and only get what you want to watch

I have my watchlist hooked up to *arr so it pulls that stuff automatically. Once I watched it and it’s not something I want to show to others, I delete it.

We use async agents heavily. The key is to have proper validation loops, tests and strong/well phrased system prompts, so the agent can quickly see if something is broken or it broke convention.

We have proper issue descriptions that go into detail what needs to be done, where the changes need to be made and why. Break epics/stories down into smaller issues that can be chopped off easily. Not really different to a normal clean project workflow really.

Now for most of the tickets we just assign them to agents, and 10 minutes later pull requests appears. The pull requests get screened with Gemini Code Assist or Copilot Agent to find obvious issues, and github actions check lint, formatting, tests, etc. This gets pushed to a separate test environment for each branch.

We review the code, test the implementation, when done, click merge. Finished.

I can focus on bigger, more complex things, while the agents fix bugs, handle small features, do refactorings and so on in the background. It's very liberating. I am convinced that most companies/environments will end up with a similar setup and this becomes the norm. There really isn't a reason why not to use async agents.

Yeah sure if you give a giant epic to an agent it will probably act out of line, but you don't really have these issues when following a proper project management flow