It appears that Anthropic has dropped the CVE/business registration requirement for individual access to the Defense Access tier. Not yet documented, and my application has not yet been approved, but X formerly Twitter users report success and the application no longer requests these details. It does require KYC via a third party.
I'm not sufficiently skilled with Photoshop to gauge how well this replaces it, but you may be interested in this supposed clean-room reimplementation of Photoshop: https://github.com/storytold/photocraft
Heard about this! Haven’t had a chance to check it out yet. I’m sure once I have an actual project that needs me to roll up my sleeves with Photoshop I’ll give it a shot. I probably should have mentioned it in my original comment as a possible option
While I like OpenCode a lot, it’s not sensible to use it for anthropic models vs. a subscription (other than to use the included API credits) in terms of cost
I sill have a ChatGPT extension option in Siri settings on 27.2; perhaps the default Apple Foundation model is Gemini-backed and the ChatGPT option still remains
Had to look this up, not great pics but very cool:
"The generator was called the Electric Power Pak. From 1969 to 1972, Ford offered a 2,500-watt underhood generator, mostly on the Contractor’s Special, though it could be special ordered on other trucks. It also wasn’t driven off the engine’s belt. It was a separate Onan generator with its own 5.5 hp air-cooled single-cylinder engine, feeding an outlet in the bed. It sat on the driver’s side splash pan. The package the commenter calls a “Rancher” is likely the Farm & Ranch Special, which is one of the trucks that came with it.
The chainsaw was real. Ford sold Ford-blue chainsaws from 1970 to about 1976. They were rebadged Ohlsson & Rice saws: the Eagle I/II and the smaller Hawk. A mount was also sold that held the saw on the F-250’s driver’s side inner fender"
This might be better than nothing for those who truly dislike Claude Code’s interface or something, but it’s basically just OpenCode as a front end to the claude application. Arguably worse for those wanting to avoid Claude’s massive system prompt, and you lose concurrent tool calls and other features. Still cheaper than the API route, so it’s something.
Mexico is more broadly a showroom for all sorts of vehicles you can't buy in the US, namely those that don't meet US safety or emissions requirements. The Hilux being the notable example, though not covered in the below link.
Emissions? These are EVs mostly. And these cars are safe enough for the EU.
Any vehicles with GPS or cellular modules made in Russia or China are explicitly banned. There's also a pending law to ban any automakers that have 15% or greater ownership by people from China, Russia, or Iran.
They are just outright banned. Has nothing to do with safety or emissions.
I don't disagree, I only mention it to clarify that "available in Mexico but not the US" isn't limited to Chinese EVs. The article's tone in my opinion implies some kind of conspiracy to deprive the US of cool EVs, when in fact there is a much broader conspiracy to deprive us of Hiluxes (Hiluxi?)!
it is deeply ironic that a toyota hilux pickup truck - an actually practical pickup truck - does not meet US "standards" but the pedestrian killing, inefficient, wasteful, and expensive "pickups" from US auto companies somehow do.
The Hilux issue is a bit more complicated as a result of the "chicken tax" of 25% on imported light-duty trucks. Toyota made the Tacoma for the US market and built it here to avoid that tax, and now the economics of the Hilux don't make sense for them to import even if it met safety/emissions standards. It would certainly kill the market for the Tacoma.
Edit: My poor phrasing certainly did read that way though, I should have been more clear.
My beloved 4Runner was made in Japan, as is the rest of them (from what I understand). One of the few models still to be made exclusively in Tahara, Aichi, that you can get in the US.
One big reason the Hilux is not imported to the US, and instead we have a separate domestically built Tacoma, is due to the longstanding Chicken Tax tariff on imported light trucks.
As the percentage of LLM-generated stuff in the world increases relative to natural stuff, I don't see how it's possible to continue to train new models indefinitely. At some point, won't LLMs be trained on a corpus that is majority LLM-generated data, with everything that entails both good and bad? Seems to me that singularity is just going to kick off a self-reinforcing spiral... Bigger minds than me have to have considered this, is this a legitimate concern?
I wonder if someone has redone the experiment of retraining LLMs on LLM output that was conducted a few years back that showed a pretty drastic degradation in quality. I imagine it's hard to perform on the closed source frontier models and the ballooning cost of training doesn't help.
In theory as the models improve their ability to ape human writing the effect should diminish, though iterative reingestion of their own output is probably still detrimental long term.
Maybe with time we'll accomplish the same thing that's happened with pre 45 steel and the instruments will need less of it. I guess in this analogy it'd be better quality filters that can detect and remove LLM-spam.
The current improvments in LLM performance are not driven by feeding even more text from the internet into them, in fact they're on the whole deliberately being trained on LLM generated text, with some reinforcement learning through a feedback mechanism (generally success on some task which requires some intelligence but is easier to verify than do). So it seems like it's not really going to be a bottleneck, though I think this RL process is part of why the frontier LLMs write so weirdly.
Broadly, the downgrading is still way too heavy-handed and clumsy. As of yesterday “How easy to crack is WPA2” downgraded to opus 4.8. Granted I could probably have phrased it differently and gotten a response, but the point is I shouldn’t have to. It can’t simultaneously be smart enough to reverse the ECDSA curve seed and not smart enough not to destroy the world by answering a trivial question with a scary word in it.
Current help doc: https://support.claude.com/en/articles/14604842-cyber-verifi...
reply