New model - that explains why for the past week/two weeks I had this feeling of 4.6 being much less "intelligent". I hope this is only some kind of paranoia and we (and investors) are not being played by the big corp. /s

▲

RivieraKid 9 hours ago | parent [-]

I don't get it. Why would they make the previous model worse before releasing an update?

▲

swader999 5 hours ago | parent | next [-]

Just guessing, but it would seem like physical hardware constraints would dictate this approach. You'd have to allocate a growing percentage of resources to the new model and scale back access/usage of the old as you role it out and test it.

▲

dminik 9 hours ago | parent | prev [-]

Why do stores increase prices before a sale?

▲

RivieraKid 8 hours ago | parent [-]

Ok, so the answer is "they make the existing model worse to make it seem that the new model is good". I'm almost certain that this is not what's going on. It's hard to make the argument that the benefits outweigh the drawbacks of such approach. It doesn't give the more market share or revenue.

	▲	dminik 6 hours ago \| parent [-]
		Tbf I don't think that it's just this one reason. While I'm not a subscriber to any LLM provider, the general feeling I get from reading comments online is that the models have a long history of getting worse over time. Of course, we don't know why, but presumably they're quantizing models or downgrading you to a weaker model transparently. Now as for why, I imagine that it's just money. Anthropic presumably just got done training Mythos and Opus 4.7. that must have cost a lot of cash. They have a lot of subscribers and users, but not enough hardware. What's a little further tweaking of the model when you've already had to dumb it down due to constraints.