Tiberium

18 ms·

361.
by Tiberium 1y ago
It does have web search - it's just not enabled by default. You can enable it with --search or in the config, then it can absolutely search, for example finding manuals/algorithms.
362.
by Tiberium 1y ago
Only an 1.7% upgrade on SWE-Bench compared to GPT-5, but 33.9 vs 51.3% on their internal code refactoring benchmark. This seems like an Opus 4.1-like upgrade, which is nice to see and means they're serious about Codex.
363.
by Tiberium 1y ago
Even V8 isn't as fast as JVM :) I think only LuaJIT is comparable in some ways.
364.
by Tiberium 1y ago
Seems like an interesting benchmark, but my takeaway from the results is that Codex is almost as good enough as their custom solution (no mention of the underlying model) and only requires a $20 ChatGPT subscription to start using it (of co
365.
Insufficiently sanitized data allows unauthenticated access to FreePBX Admin(labs.watchtowr.com)
52 points by Tiberium 1y ago|10 comments
366.
by Tiberium 1y ago
It's sad that HN top nowadays has articles that are completely AI-edited or AI-generated, with the default style that's obvious to anyone who has used them for enough time.
367.
by Tiberium 1y ago
It'd be curious to see how those AI generated kernels compare to kernels generated by https://github.com/tinygrad/tinygrad
368.
by Tiberium 1y ago
The article itself appears to be largely AI-edited. And I'm really surprised that anyone would want to write an article on this, I assumed it was widely known? You can go onto Censys and find thousands of exposed instances for lots of
369.
A Review of Nim 2: The Good and Bad with Example Code(miguel-martin.com)
2 points by Tiberium 1y ago|1 comments
370.
by Tiberium 1y ago
Found out from some Reddit discussions that the developers aim to first upstream everything for M1/M2 to the kernel, and as of https://asahilinux.org/2025/08/progress-report-6-16/ : > With Linux 6.16,
371.
by Tiberium 1y ago
Sorry to hijack, but since the topic is related: is the development of Asahi Linux still actively ongoing, or has slowed down a lot? The progress for M1 and M2 was steady and now almost everything is done, but the M3+ work still seems to no
372.
by Tiberium 1y ago
The whole README is heavily AI-edited (the final output is all by AI), and the worst thing is that the image diagrams seem to be generated (likely with 4o), see for example https://github.com/MaliosDark/wifi-3d-fusion&#
373.
by Tiberium 1y ago
I do agree that Apple Silicon is way, way more energy efficient and comparable to some top desktop x86 chips, but.. From reading different reviews and tests apparently it's normal for the CPU/SoC to reach a temperature of up 100C
374.
by Tiberium 1y ago
You can, but zigcc is essentially Clang bundled with extra stuff to make cross compilation easy.
375.
by Tiberium 1y ago
Can you provide some examples if you still remember?
376.
by Tiberium 1y ago
It's not a built-in Nim feature, and it'll only work with native backends (C/C++/ObjC). The project that makes it possible is https://github.com/yglukhov/nimpy
377.
by Tiberium 1y ago
Yes, compile-time execution in Nim is very natural, if you mark a variable "const" or use some other ways, you can run almost all pure-Nim code at compile-time without having to modify it.
378.
by Tiberium 1y ago
I use LLMs almost daily, and they certainly don't "solve" the problem of finding bugs, not even in popular languages like TypeScript. They do help a lot, yes, but they haven't solved it completely.
379.
by Tiberium 1y ago
The first character is actually not considered for case insensitivity, although it was in the past. :) And case insensitivity itself comes to Nim from Pascal.
380.
by Tiberium 1y ago
I think Nim and Zig target very different audiences and have very different goals. Nim is about being a big, powerful language with lots of features, so that you have the freedom to use it the way you want, e.g. there is OOP with methods, b
381.
by Tiberium 1y ago
HN doesn't have DMs :)
382.
by Tiberium 1y ago
Opus 4.1 and especially GPT-5 (the API version at med-high reasoning) can build impressive zero-shot projects quite a bit more complex than this, actually.
383.
by Tiberium 1y ago
I must say, this is an interesting way to advertise the AI service this was made with.
384.
by Tiberium 1y ago
Seems to not be the case, a real 51% attack would need 10 blocks at the very least, because that's when Monero transactions get confirmed. See e.g. https://x.com/kayabaNerve/status/1955173552363016434 https
385.
by Tiberium 1y ago
It's not, just like o3-pro is not the same as o3 with high reasoning.
386.
by Tiberium 1y ago
> However, to be able to set the reasoning effort you have to switch to the new Response API, which wasn't a lot of work, but more than just changing a URL. That's not true - you can switch reasoning effort in the Chat Completi
387.
by Tiberium 1y ago
The $200 sub (ChatGPT Pro) offers GPT-5 Pro, which is not the same model.
388.
by Tiberium 1y ago
GPT-5 in the API, especially at "high" reasoning is quite a bit better than o3, especially at web design "style". Also from my own experience, and from some people I've talked with it's great at agentic program
389.
by Tiberium 1y ago
Respectfully, I'd have to disagree, the models are extremely filtered, it seems like they have almost no real world knowledge and are likely trained directly on o3/o4 mini outputs (synthetic data, like the Phi series). Even then,
390.
by Tiberium 1y ago
V8 is extremely good, but (maybe due to JS itself?) it still falls short of LuaJIT and even JVM performance. Although at least for JVM it takes way longer to warm up than the other two.

More ›