You know they are benchmaxxing when they end up writing their coding harness in TypeScript npm slop

Their models can't help them build it with something better?

That's the only benchmark people need, whether or not their model can raise the bar of their own product

And so far it's looking pretty sad

[dead]

[dead]

[flagged]

[dead]