The few days with Mythos/Fable were magical.
This whole situation has shifted my perspective on the importance of open source / open weight models, and the impact of AI this powerful overall.
@kimmonismus Grok 3 is the better product rn
Claude 3.7 is the most useful model yet restricted
OpenAI is the best hype machine. Casuals will always enjoy it.
IMHO, the key has been alignment… Claude’s competitive edge is due to their work in trying to understand the model from the inside
@DaveShapi What do you make of the “jailbreaks” then?
Hallucination? Roleplay?
Either way the models still show radically unaligned behavior in these scenarios.
As someone who’s had similar experiences to yours with Claude and O1, I’m curious of your take on this.
We are launching a $99/year project management tool for 100% FREE for limited period of time at @PitchGroundHQ
Comment “pitchground” and RT to get access for free when it comes out on 24th Nov!