I hear the argument here, but isn't it possible it has dramatically more knowledge and when you get outside the common cases many of us use it for, it'll have completely different capabilities?

I feel like most benchmarks cluster on a reasonably limited area of human knowledge

Sort of depends on how well the core reasoning works. It’s not a big effort to connect an LLM to a search provider.

You do pay for the tokens, but in theory on a smaller model each token is cheaper.

honestly using search isn't that great, you mostly get SEO slop, it usually won't help the model ask the right questions

When I messed with it I used Kagi's search and I didn't have that issue (not claiming they're the best, they're the only one I tried).

They filter their results through their AI, though, so you get a sort of meta-summary of the top few results. It did well with geopolitical news stuff, but I've not tried a hard science sort of query.

Try Parallel.ai (no affiliation). Instead of keywords, the model writes objectives.

It seems like there is tradeoff between model size and the need for tool use, which - in my mind - is quite costly in terms of time and tokens. More detailed world knowledge requires an exponential increase in model size, but most knowledge can be acquired ad hoc using search or database queries. This will fail for questions where the model lacks the knowledge to ask the right questions, but maybe this could be solved by a handful small inquiry models with different knowledge encoded in their weights?

Or calling into a full-knowledge model “I’m facing problem x, how do I ask myself the right questions?”

I should do that myself, come think of it.

I think the Western labs are burning through funding and compute to maintain the lead at any cost, efficiency be damned.

You can't compare models released 6+ months apart. GLM 5.2 was same architecture as 5.3 and not nearly as good. Takes time to build frontier intelligence and distill down to smaller sizes.

It's not that GLM5.3 in full precision unquantized is any smaller, it's 141 * 5.4GB files at approx 770GB which is about the same size as 5.2.

Hold on.. the routed experts are in FP8 now? Previously they were in BF16. Nice, this shall cut my download time by half!

This time they just made FP8 "default", accompanied by "-BF16" model/page (previously "-FP8" was released alongside).

4.7 and 4.8 are smaller than 4.5 and 4.6

(And 4.5 was smaller than 4)

I am curious why you say that?

Not that I don't doubt the veracity of it.

Just curious if you have any proof?

> that would prove rather embarrassing for Anthropic

Not really, in that you just work with different constraints.

Anthropic and US labs in general has maybe 100s to 1000s of GPUs per person to experiment. Zai and Chinese labs in general have 1-10.

The priorities are different.

And the Chinese labs still make models that are easily as good as the US labs.

Rather embarrassing indeed.