My experience after one day:
Code completion seems not to work - at least not as well as Junie does.
But the quota is really good. I used it all day and dashboard says 16% is gone. AI Pro Quota would be empty by now.
So the amount of quota justifies $30.
k
Kevin
09/14/2025, 5:44 PM
also not all quota is the same in value. the firebender coding agent performance is pretty far ahead compared to junie in both efficiency, speed, accuracy
its like giving a middleschooler a hammer and a professional carpenter a hammer. you're going to get very different looking homes
s
Stefan Oltmann
09/14/2025, 6:39 PM
I wonder: If I define GPT5 to be used for both, should there really be a difference? 🤔
k
Kevin
09/14/2025, 7:08 PM
gpt 5 is the hammer. maybe call it a toolbox even
Kevin
09/14/2025, 7:09 PM
coding agent performance is quite different even though they use the same model
ie:
claude code vs aider
cursor vs copilot