rockdoe
rockdoe on Hacker News
Due the the size of the default system prompt you get over 100k really in easy in Claude, and because that is fixed 200k is really almost 3x more room to work with, not 2x.
200k also used to be the limit for Sonnet for a long time.
If Haiku 5.5 also had 200k, it'd actually be usable for us. As it is, it's Luna or Sol.
In Claude it used to be the circle on the lower bottom, but now it's a bit more hidden and you have to expand the sidebar.
This isn't even a good attempt at trolling, do better.
Luna and Sol-6.1 are crazy cost effective and all flows here have switched to them, so I'd say really well so far.
There's very few stuff that needs Astra/Opus level intelligence.
Luna 6.1 would be amazing.
the surprisingly capable Haiku 5.5Destroyed by Anthropics crazy decision to price it x5 after 100k context.
The "realistic setup" used for this didn't have a php page on it initially.
I'm skeptical of the value of such setups, unless it's clear it's an out of the box setup, and we already know those are totally broken.
I don't understand who these people think they are really fooling. Is this some AI denial?
It took about 2 hours to get taken over and that was with a setup that's sufficiently non-standard that existing PoC would've needed tweaking.
The fact that people are still in denial over this is crazy.
I have working RCE as long as the server has a memory_limit that is less than the max POST size. From my reading of the PHP patch, there's no way around that part, but a lot of real sites are running like this.
No idea of course if this meets the authors' "out of the box" condition. Who runs a fully stock PHP config?
I don't think it's implied the wrapping keys were the weak ones. More like, the script that ran the wrapping still had the fast "debug" generation of a pair of keys to test the wrapper on enabled, for example.
I have no idea why you wouldn't be using Qwen3.6 35B-A3B with that setup. Or did you mistakenly think that was what they're talking about?
Their model is over a year old, like, a few decades in LLM history.
If your generation is any good the odds of ending up on such a number should be astronomically small.
Biasing something that should be random tends to badly backfire. See the Enigma who avoided enciphering letters to themselves, which was then used to break it.
So uh, for various IOT junk like Smart TVs, does that mean the ones that don't get security updates (pretty much all of them) are permanently pwned?
Native DLP is specifically listed as a feature FYI, also mentions compatibility with Broadcom and Trellix.
Edit: "website filtering" too
It took a while to point out all the other stupidity in the contract (this wasn't the only part, just the most blatantly silly) and I ended up taking another one.
AFAIK Firefox add-ons go through review before being uploaded, so this would actually only be an issue if you don't trust both the Canadian guy and Mozilla.
And it seems like you cannot apply new policy while the browser is running - meaning it cannot be used to roll out hotfixes without interrupting the work of currently unaffected workstations.The changelog linked above seems to hint this is specifically fixed/a new feature in the Enterprise version:
"Many policy-driven changes can take effect without restarting the browser."
I don't know, but it's a good question? I have a bunch of test questions that Claude is unable to solve using the web, any version, Fable 5.1 to Haiku 5.5, and pretty much every OpenAI model gets it easily.
My only explanation is that Anthropic's search backend sucks.
MSIE style internet globe or GTFO
I've been doing consultancy for OSS development and one in-between company asked me to sign a contract that had a random clause in it somewhere that banned the use of OSS for work.
When I pointed this out, they said "just ignore that part".
Nope LMAO!
I mean I get it, as an admin there's only so much time you want to spend to support everyone's favorite setup. So "security/safety" is kind of a simple wave-off. It's sometimes just "we don't want to spend the time to figure out how to lock down this thing."
Someone dropped the following link, which seems to have way more detail than the press releases:
https://www.firefox.com/en-US/firefox/enterprise/158.0/relea...It looks like they do actually offer that as a feature: "Employees can access internal company sites and web apps from personal devices without a device-level VPN"
Edit: To give proper credit, the link is from Heise.de
It doesn't need rights to overwrite system binaries, it just needs enough permissions to tickle the updater. And the updater has the rights, for obvious reasons.
Pretty sure that's Firefox in contrast to the other alternatives in the Enterprise space.
Do you have a link to the bug?
BTW. The "won't load new pages" may be something else. If an agent launches its own Firefox in a new profile, it will sometimes trigger the auto-update which will overwrite the binary in the profile you were using. It won't crash but you'll get a warning to restart.
> He didn't even proof the last theorem of Fermat directly, he proved some conjeture that someone else before him, proved that it implied Fermat last theorem...
That's how maths works yes...
Of course they are "reading user prompts" in the sense that it's being used to further training. That's why you have to pay up to opt out of that.
That's also why they refused to answer that question: the answer is obviously "obviously"!
How many of those are false positives though? Probably just over 5000?
You get bug bounties if you report the kind of bugs Mythos identified. There's a reason no-one collected bounties from the "5000 defects" Coverity identified.
The Mythos reports have several examples of chaining a whole bunch of logic in different parts of the program together to exploit something very subtle. The Coverity reports aren't anything like that. These tools aren't remotely in the same league or even universe.