GPT 4.1 performs better than Sonnet-3.7 for programming tasks involving long contexts

I have been doing analysis, planning and debugging using GPT 4.1 a lot lately, in agent mode, he is fast. I also have a second chat tab open for 3.7-Sonnet thinking mode to do the nuts and bolts programming by passing a temporary generated Cursor rule to codify the ideas and transmit to 3.7-thinking to deploy … Then do a code walk through and critque using Grok-3-beta feeding that back into 3.7-Sonnet again

I even have a rule about how to create rules

1 Like