Are we sure this is a new pretrain and not them rounding up 2.8T?

vikramkr 6 hours ago | parent | next [-]

Literally from the linked announcement:

> Kimi K3.1 Pro is a post-trained version of Kimi K3, with substantially larger post-training compute

minimaxir 6 hours ago | parent | prev | next [-]

The Kimi K3 announcement (https://www.kimi.ai/blog/kimi-k3) used 2.8T and it would be weird to round that up for this announcement.

	▲	Philpax 5 hours ago \| parent [-]
		Looks like this post's using "3T-class", which I think I've seen them use before.

rfoo 6 hours ago | parent | prev [-]

it's almost certainly rounding up 2.8T