There is literally no evidence behind these models which would indicate that they are distilled.
Same things were said about Kimi k3, and yet it scores much higher in certain areas than fable. Thats not how distilling works. If distilled it would score the same or worse. Not higher.
An additional factor is cost. The model the model cost more than 100x less than fable. It's cents on the dollar. It's a completely different architecture.
Like genuinely you brain dead ai monks needs to find a new area to post on.
@vansh22b X user: "Codex limits are still shit"
Actual codex user on 200$ plan: "What plan are you on"
X user: "20 dollar plan"
Actual codex user on 200$ plan: π
@arjunkshah21@ycombinator I dont mean to be a hater but you can jut put codex on a loop with a goal and it will build this same thing with 4 hours or less.
New tech used to be fun, now I just know it was made by a robot in a few seconds.