Julio Pereyra
@ItsJulioPereyra
Kimi is a cool model, but I think this graph shows the opposite of Will's argument re: our work.
That a model that was likely post-trained on LAB (Kimi K3) handily beats out a smarter model (by AA overall intelligence) that was not post-trained on LAB (Fable 5) shows exactly how
That a model that was likely post-trained on LAB (Kimi K3) handily beats out a smarter model (by AA overall intelligence) that was not post-trained on LAB (Fable 5) shows exactly how
Will Chen@willchen500 · Jul 18This is the Deepseek moment again but even crazier. OpenAI and Anthropic IPOs might be derailed if this keeps going on.
Also this proves how pointless Harvey’s second try at post-training is. They have again burnt a ton of investor money for nothing.
They should just host the
Open quoted post →Also this proves how pointless Harvey’s second try at post-training is. They have again burnt a ton of investor money for nothing.
They should just host the
5 25