Hacker News Reader

3.5 vs 4 is like day and night. Try to get you impressions on v4

I must have a play with GPT4 at some point. For me, 3.5 generates working code more often than not and rarely hallucinates.

Not the OP but good to know. How does 4 compare to Bard?

3.5 is arguably better than Bard on some coding use cases and higher accuracy. 4 is just much, much better than both of them.