v4 flash has been working quite well for the majority of my personal projects, with occasional v4 pro or Kimi 3 for the most complicated tasks or to check the overall project progress (when vibe coding).
I must be doing something wrong. I gave v4 pro a try a couple of days ago, gave it a simple prompt like "clean up functions x and y in file z" and it would always start off promising, just to quickly get sidetracked, start hallucinating problems in the code, and just get stuck for hours until I interrupt it:
— hmm — 0x2D696370 — little-endian bytes: 70 63 69 2D = 'p','c','i','-' — hmm — WAIT — WAIT — !!!!! — *WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — *HOLD ON — HOLD ON — HOLD ON — WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — !!!!!!!! — *WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — *OK — WAIT — I THINK I FINALLY SEE THE WHOLE PICTURE — I NEVER READ IT — AND — THE LAYOUT — hmm — !!!!! — *WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — WAIT — HOLD ON — HOLD ON — HOLD ON — HOLD ON
Then gave the same to Sonnet 5 and it was done 15 - 30 minutes later.
I tried v4 pro both in claude code and codewhale with similar results. Haven't tried the new deepseek harness.
I've been using the 0731 Flash V4 model, via Opencode, and I've had no major issues. It feels very comparable to Opus 4.6/4.7, that I use at work. I haven't ran into any of the problem you mention, so that might be a quirk of V4 pro, the specific harness, or maybe the host you're using?
I guess this was a heavily quantized version from openrouter? I've never had that experience in the last months of quite intensive use of the official deepseek api.
No, that was directly from deepseek. And it happened very consistently (3 or 4 times in a row, clean session every time, and across 2 harnesses). Guess I'll give it another try when v4.1 flash is available.
It's a bit depressing. I personally try to enjoy the present with my loved ones as much as I can, the future is a bit impossible to forecast right now. I'm focusing on staying alive, mortgage payments, etc.
If I was a company with a zero data retention contract involving OAI I would be asking for a third party audit of such claim of zero retention like, yesterday.
By the way, the company that made it's entire product off of stealing all data it could get it's hand on while violating copyright and pirating, is not all of a sudden going to respect your data. If you think OpenAI or any major AI lab is going to give you true ZDR, I have a bridge to sell you.
So use bedrock or vertex or whatever. Those are the ZDR offerings. Or was it your intention to insinuate that the major cloud providers are conspiring with openai to violate their contractual obligations to their customers?
Yes. You're naive if you think any of these cloud providers care about your data when they're all in the midst of a AI revolution psychosis. They dont care about their reputation or what you think of them, they think they're going to have a machine god their side.
If my company finds any evidence of OpenAI violating ZDR, we'll sue for breach of contract and fraud, and collect damages. I think we'll be able to afford the bridge you're selling. You've got the title and title insurance, right?
I'm certain my company didn't agree to arbitration, big bro. Our lawyers are putting the fries in the bag.
Isn't OpenAI being sued by Apple for their little stunt?
If you're saying "the big bad guys always win anon, just take the black pill," then there are tons of counterexamples. Remember Uber paying Google a sweet Bil for pulling this same trick with LIDAR firmware?
This is how every conspiracy theorist thinks: my enemy is Bad, and if they did a Bad thing, it would be Good for them, therefore they obviously did it. No evidence needed other than "motive" + my enemy is evil. But even if your enemy is evil, in this case, they would be fools to take the legal risk of violating their contract for the minimal upside of a tiny bit more training data (and fools to assume this would not be exposed in a large organization). So you need to assume your enemy is both evil and remarkably stupid.
I think it’s probably not surprising that they would go up to the contractual limit or into a grey area; but exceeding that would require too much coordination among individuals, as you say.
> (the goal here was not to scoop any particular individuals and we were looking at many problems beyond these)
That is your opinion, but the optics of that should raise for you some flags. OAI could have waited (how long is a task left to the ethics committee) to see how the rumors panned out. Right now the optics look a lot like "we don´t care there is a 1/7 chance we one-up a human researcher by reacting to this rumor immediately, might makes right"
> "When we learned that they had Euler but not Navier-Stokes, we offered to let them go first, to suggest that they should be the ones to get the prize, and optionally for Tristan to be the lead author on a rewrite of the OpenAI proof. We felt it was challenging to offer the same to Levent (an Anthropic employee), who was not willing to talk or coordinate with us anyway. We were open to other solutions."
What an admission! "We tried to defraud Alpöge out of sharing the Millenium Prize (that we don't dispute he might actually deserve), for no other reason than he works for our competitor and that inconveniences us".
I thought Tristan Buckmaster's allegations sounded fantastic; and then 'sama just came out (tweet's ~30 minutes old) and admitted to all of them. Wow!
That isn't what the quoted passage says though? The claim by openai (no idea if true) is that they offered to wait for the other two to claim the prize before publishing their own work. Separately, they also offered to let one of the pair (but not the other) become an author on their own separate work.
Interesting that they quote the mathematician directly: “there is nothing you can do, I simply do not trust you”
but then they proceed to NOT quote themselves themselves verbatim: "I deeply apologize for this extremely poor choice of words, it is the opposite of what I was trying to convey."
Talking like that and threatening an academic like that is crazy. I read the explanations Altman and the others posted and they completely skip over the whole "I don't have to be nice" style threats.
That's what distinguishes super yachts from mere regular-sized yachts. The wide base of the hull increases stability on the water to the point that donkey husbandry becomes practical.
Or maybe it is just a legal way to escape taxes. Enough to be worth owning the yatch and paying for its crew and maintenance. Just that should give a hint to the low bar of the money amount involved.
Everything is high tech except for the cat litter collection, which tells me the owner already knows cats don't like automated litter collection. I wonder how hard is to keep the house clean tho!
It would require an authorized human to load the floppy into the drive and make it go. And there's nothing special about that process being done in code, it could just as well be done by people flipping switches manually.
There's just no plausible way, in the real world, that a computer program running in some datacenter is going to make this happen. Or anything of any dire existential consequence. It's not a real threat.
I think I disagree! You can always manipulate a human. It's true tho that this particular example has a myriad of failsafes, so the actual example I wrote is less than interesting.
Yes, but to make "apocalyptic" scenarios happen your AI would need to manipulate at least millions of people. So far it's not yet clear they can successfully manipulate even one, or even that "manipulate" is a coherent concept, since LLMs pretty clearly have no agency or intrinsic goals. So like... I guess if we start seeing AI systems autonomously throwing elections and starting wars I'd be more concerned? It doesn't look likely anytime soon.
As a consumer I feel like hansel and gretel combined, deepseek could be the witch.
reply