Comment by rafram
5 hours ago
4.6 was released at the beginning of February, so if the Chinese models only "tickle its tail," that means they're >5 months behind.
5 hours ago
4.6 was released at the beginning of February, so if the Chinese models only "tickle its tail," that means they're >5 months behind.
That comparison is also misleading because Opus 4.6 was probably not Anthropic's frontier model.
We got the first news about Mythos in March, so it is likely that it was already close to ready by the time Opus 4.6 was released.
So the actual gap is the time elapsed between March (or April for the official announcement) and whenever Chinese models can match Mythos.
The post-training process of a model that size is months, though it "works" before that. It is a big chunky model before it's released to the world and probably does some amazing things, sometimes...but, it wasn't done (else why wouldn't they release it and soundly trounce their competitors). I would assume that Chinese AI companies have a pipeline and what we see is a couple/few months behind their newest model, as well. Like, the new base model is cooked, but they're still plating it for service.
Why would Anthropic get the benefit of pre-release models counting toward their lead, if nobody else gets to count their pre-release models?
> The leading Chinese models are only a few months behind now
I hear that often, but what does that even mean? I am a great proponent of open weights models. I do believe they are the only reason we have not stagnated into a collusion of halting (public) model releases.
But exactly which point in time is z.ai compared to claude.ai? Consistently bring "6 months behind" in an exponentially acellerating evolution means the gap is growing exponentially wider, not constant.
What range of numbers do you believe "a few" represents?
Opinions vary, but:
A couple: usually 2, though not always
A few: 3, 4, 5
Several: 4, 5, 6, or 7.
Several and a few are the same number, they only differ rhetorically.
> A couple: usually 2, though not always
I had to explain this to my German friend. In my understanding this isn't about the actual number, it's about the certainty. If it's absolutely and definitely two, then I say two. If I'm uncertain but it's probably two, or if a non-integer, somewhere around two, then I say couple.
And few is more likely to be 3 than 5, because 5 is getting close to a "half-dozen or so", or (as you say) several.
Many is very context-sensitive, as the meme has it.
So I would agree that the open models are a few months behind, definitely more than a couple of months behind, possibly several months behind, maybe a half-dozen months or so behind, but not many months behind.
1 reply →