I really wanted to believe in this free AI coding tool could replace Claude Code. But it isn't ready for prime time unless you're willing to babysit.
On a 2.0 terminal benchmark, OpenAI’s model scores about 10% higher, guiding users toward stronger results on long, complex ...