Press releases are sadly like that. Fujitsu's main page has it in the first line:
> FUJITSU-MONAKA is a Next-Gen Arm-based processor, set for release in 2027, designed to address the challenges of next-generation data centers with its unmatched performance and power efficiency.
Qwen 3.8 is excellent. With the right harness, it does about 95% of what Opus can do, in my case automation software development. Since 3.8 came out, I have significantly revised my expectations for a local model. Give it another year or two, and we'll be running fast and free local models for nearly everything that matters, and these costly subscriptions will be a thing of the past. I've always believed that AI should be free for everybody, like TV and radio. We're almost there.
Free tv and radio? Where do you live? Where I live you either pay taxes for it, alternatively, it is so ad infested that it is not possible to watch it.
I suspect the same will/is happening with AI. Either you will pay for it, or it will be so ad infested that it will become useless.
Not op, but check out open code; you can turn on K/V quantization to help with increasing context if you have not already. I think K needs to stay at least 8 but I hear V can go down to 4?
I am running it on 32GB and I did not saw model loosing it context even after 4-5 compactions in pi. I am running sessions for few days sometimes. I think it looped once, but loop police extension stopped it. The only problem I have know is how pi compaction works, which is forcing full prefill which takes time and it is erroring a lot. I wrote my own compaction that should remove full prefil but it does not work. But this is the only problem with this setup and it is more problem with pi then the model. I much more prefer it to use Qwen then paid models: Claude forces me to do reauth every other day and codex models either are too costly or not capable enough.
I've a 32gb and 64gb (work) MBP. 32 works - just and sits at around 28/29gb of 32. 64 works great, so the 48gb laptop with MLX + MTP should be fine. I'm using Ollama.
I initially used the Claude Code harness on 3.6 A3B, but found that tooling would break as Claude released new versions and things would go weird. I've since written my own harness which has basic operations: read, find, bash (which can write files, python etc...) & web_fetch, all within a mac container. Works amazing. You don't need anything complicated to go very far.
Low hanging fruit would be Pi or OpenCode. If you really want a much better understanding of what your hardware is capable of then give writing your own a go.
Additional tip: Low Power mode reduces some token speed, but stops the laptop over heating and the fans going crazy.
So, pay for access, or go to your local library, or browse the internet, or get free access in a similar way that Aaron Swartz did? Knowledge often requires funding to be created in the first place.
I mean, they don’t, exclusively, or no-one would be giving money to JSTOR. And certainly, no-one would be scanning millions of old journal articles for free.
I don't know where people get the idea that JSTOR is one of the bad guys. It's a non-profit that scans old journal articles that would otherwise be a huge pain to access. It costs money to do that. What specifically have they done that you disapprove of?
Beats me what you're talking about, then. The literal answer to your original question is that people use SciHub and Library Genesis because they're free. But they're free because they infringe copyright. Maybe Swartz was opposed to copyright per se or maybe he was trying to make a point about free availability of journal articles associated with publicly funded research. Either way, JSTOR is the wrong target. JSTOR doesn't hold the copyright on those articles. JSTOR does actual legit work to make scientific articles easier to access – which is exactly what Swartz seemed to be in favor of.
JSTOR provides digital access to print journal archives. That's useful to anyone who wants to read an old journal article without going to a physical library. I used to have to go to the library, dig out the relevant issue of Linguistic Inquiry, and photocopy it. Then in 2015, JSTOR started providing access to the entire scanned Linguistic Inquiry archive, so I could access all of those issues digitally.
I don't understand what you mean by "money-clutching noise" in your other comment. I think enabling free universal public access to historical journal articles would be a great idea. However, some public money would then need to go towards the costs of that, and some reform of copyright law might also be needed.
Practically speaking, no-one is going to scan centuries' worth of historical journal articles for free. And JSTOR isn’t really an entity outside the research community; it’s a non-profit that grew from within it.
I think we’re talking past each other mostly because I start to lose value on most research for my topics of interest that are more than two or three decades old, let alone five or six.
> [JSTOR is] a non-profit that scans old journal articles that would otherwise be a huge pain to access
You'd think that would pretty obviously include the vast swathes of material published before the internet and PDF preprints were a thing (and the considerable amount of subsequent research that just never got uploaded to anyone's website, for whatever reason).
Anna's Archive is primarily a meta search engine. I see that they've put out a call for volunteers, but I don't think any significant part of their archive consists of volunteer scans.
As to your 'why' question, we're talking about boring, thankless work that in many cases breaks the law. It's not exactly surprising that we don't see millions of people signing up to do it for free!
At present I don’t see individual contributors making any significant contribution towards scanning old academic journals. Why would this change in the future? There was at least a decade where the technology to enable this existed, and where most older issues of most journals were not available online, and yet the army of volunteer scanners failed to materialize. Now that there is already a non-profit doing this archival work at very high quality and with economies of scale, it seems even less likely that such a volunteer effort will materialize. But if you want to prove me wrong, get off HN and go scan some old journal articles!
SciHub mainly redistributes scans of print archives made by JSTOR and other organisations. And it does not have everything that JSTOR has. For example, this paper is not, as far as I can tell, available on SciHub:
Somehow, you appreciate SciHub without appreciating one of the organizations that did the actual work necessary to make many of those PDFs exist in the first place:
This looks like it would be cool, but it is quite broken.
reply