@depatchedmode
This latest generation of models has been a *substantial* upgrade. They are now pushing back on things and arguing for the correct path forward (understanding the context you've provided) even without being prompted to be disagreeable.
And the agent-based coding results are huuuuuuugely improved from even a couple of months ago, particularly when negotiating a plan in advance of executing anything.
0 reply
0 recast
1 reaction