The mac mini is a wonder but it's not a great gaming machine[1]. You can see that these stats are about 1/2 of what the Steam Machine does, so I think the comparison is pretty apt.
Neither of those options is good value. PS5 charges more for games and you have to pay extra to connect to the Internet. The value proposition of PC gaming is openness. You can play on whatever you want, on anything that can run it; you're not locked into one hardware vendor or game store. This means competitive pricing across all market segments, except where consoles sell at a loss to buy lock-in.
Buying a gaming pc is always a bad deal compared to a PS5! Even though anyone buying a gaming PC is getting a "bad deal" - many people prefer it. You can do lots of things on a PC that you can't on a PS5 - and there are reasons someone might want a 6" cube instead of a full PS5 and a mac mini. None of them are low price but they are reasons nonetheless.
A great example of the target audience are the people who've been playing games on the Steam Deck, but want something with a bit more oomf without the hassle of building a PC. I am not in that demographic! But I have a friend who is. He's quite happy to pay more for convenience. He already has a gaming laptop, but I can see him getting this to replace his ancient Steam Link.
> Buying a gaming pc is always a bad deal compared to a PS5!
If you only compare the hardware, that's true. Even if you don't consider all the other functionality that a PC has vs. a console, add all the different ways to get free and heavily discounted games on Steam/PC, and the results of that calculation might start to look very different.
Your response is essentially OPs reasoning, read it again :)
Anyways, just wanted to add that the steam machine and PCs killer differentiator: a truly open platform that no mac, ps5 and other consoles can offer. Do whatever you want, install whatever software you want, whatever OS you want. Break the rules, face the consequences. Live life like a living being, not as a slave to some corpo.
I take it you meant GP (as in, the post I was responding to - which to this post is actually GGP but I digress).
I don't think it is. Their reasoning is:
> there are reasons someone might want a 6" cube instead of a full PS5 and a mac mini. None of them are low price but they are reasons nonetheless.
Mine is that it is indeed price, only not the price of the hardware alone but rather the price of the ecosystem as a whole. Another aspect that I didn't cover is that a game that you buy today for PC will likely still work on whatever PC you have 20, 30 years from now. The same cannot be said for consoles.
what makes a great gaming machine? It plays Dwarf Fortress amazingly. And nethack. I'm running factorio on my Mac. The older Mac mini's run windows. Game of Thrones a ton of money at proton so anything that runs Linus and X86 has a shit ton of games yeah even if it doesn't have your pet game.
A piece of hardware that runs a basket of popular higher-end games at close to 60fps is generally what people look for. If you know you wanna run DF you can use much cheaper hardware, but if you wanna run "games" you wanna check that your target pc performs good enough on a selection of games.
The lowest spec M4 Mac Mini on apple.com is $799 today. The next generation Mini will likely be more expensive due to memory pricing, and as the Steam Machine already includes current higher memory pricing, that would be a fairer comparison, no?
Maybe in the future. There should be a new generation of Mac Mini's soon, further extending the performance lead of Apple chips.
Maybe once Fable is back or the next OpenAI model releases, we could take a look at implementing a compatibility layer to translate DirectX games to Metal.
Even if that should yet be out of reach, such a project may become more feasible if AI progress keeps up.
I agree. And at work it has been producing some of the worst GUI test cases I have ever seen.
What is tested often makes no sense at all, completely implausible edge cases are tested on internals, while it doesn't create tests for the overall application using user events.
And some things in these test cases are downright ridiculous: instead of instantiating your classes, it sets up some barebones fake objects reimplementing some of the behavior of your actual class, then ignores the TypeScript errors via force cast or similar.
Then it proceeds to slap some test ids on the output, stubs components and dependencies more or less randomly, adds some assertions on test ids and calls it a day.
Apparently that's good enough for many colleagues to open a MR for that garbage.
That said, at home with SOTA models I happily hand large units of work to it, outsource much of the thinking, and get workable results. I think this is the future.
I previously gave this author and the bun rewrite the benefit of the doubt. But an obvious slop PR to the WebKit repository?
I'd tap out here too if I was a maintainer. Even if the change was perfect, if you could not be bothered to write the PR description, I am not going to waste my time with it.
Edit: My bad, the PR is to a fork, in that case it's not our business how the PR description is written.
GitHub Copilot in vscode has two ways to access Opus: the Copilot harness or the Claude Code Agent SDK within Copilot.
And that's if we assume that the vscode GHCP default Agent ("Local") is the same as the "Copilot CLI" one that is also selectable in vscode. I have not tried that one.
A few weeks ago the Claude Code Agent SDK was much better than the default Copilot Agent, but nowadays I am not sure.
Has there been any evidence of a well known provider rerouting to lower quality models?
Last I saw, engineers working at OpenAI denied this on HN.
I saw that someone set up a tracker that aims to record the performance of the models, and so far it has not shown any statistically significant deviation in performance for Codex, and not yet enough data for Claude: https://marginlab.ai/trackers/codex/
Model re-routing happens for coding tasks too. For example, in OpenAI support pages used to (at least 1 month ago when I checked) mention that if they automatically use a cheaper -mini to accomplish the task behind the scenes, you’ll be charged -mini prices even if you selected a more expensive model. I just checked again and they’ve removed it, but there’s probably archives.
Finally, even if they’re the same weights, you don’t know what quantisation you’re running at. Adaptive quantisation based on load (given workday peaks), or similar techniques, have been happening since the ChatGPT 3.5 days; the techniques are probably more advanced now.
What you linked appears to be related to GPT-5's Auto router in the ChatGPT app back then, and that it supposedly would choose the 'good' model over the pretty bad Instant model for mental health requests.
That's pretty far from the hypothesis that either OpenAI or Anthropic is using adaptive quantisation based on load for their professional coding agent tools.
This is what I think engineers working at OpenAI explicitly denied, and for which we have seen zero evidence yet.
Many people seem to believe it anyway, but the non-deterministic nature of the tools appears to be the more plausible explanation for perceived degradation, in my opinion.
There are numerous comments here from experienced people addressing this. Yes, that happens and a doctor who dismisses the concern can be sued for malpractice if something actually does show up, so they are put in a difficult position. For some reason you just assume that doctors will recommend against an invasive procedure when there is a positive tomography result.
Review the numerous comments that address this as a statistical issue -- which it very much is when talking about the scale that Midjourney is claiming.
However, their "overbuying" of compute means they can now rent it out for $2.32B/month.
That seems like solid business to me, and raises the question whether your claim of xAI needing 'ongoing enormous funding' is accurate.
reply