it’s a matter of metrics, sure, but vibe-level it would be hard to convince me that all this knowledge and innovation has led to decay
August 22, 2024 at 2:05 PM
it’s a matter of metrics, sure, but vibe-level it would be hard to convince me that all this knowledge and innovation has led to decay
C3-Opus seems to continue to be the most empathetic model, with a mild trend towards aesthetics and flair. I think, on the whole, that I like it the best of the four, but the tooling around it is less usable than GPT-4(o).
June 22, 2024 at 4:54 PM
C3-Opus seems to continue to be the most empathetic model, with a mild trend towards aesthetics and flair. I think, on the whole, that I like it the best of the four, but the tooling around it is less usable than GPT-4(o).
GPT-4 is more likely to intuit what I mean by my questions and follow instructions, and has sometimes been more accurate. GPT-4o, however, is better at shutting up about qualifying its answers and getting to the point.
June 22, 2024 at 4:52 PM
GPT-4 is more likely to intuit what I mean by my questions and follow instructions, and has sometimes been more accurate. GPT-4o, however, is better at shutting up about qualifying its answers and getting to the point.
I keep asking them various questions to try and pick up on the different vibes, but honestly... I can't tell. GPT-4o and 3.5 Sonnet feel more likely to use formatting, and that's about the only consistent difference I can see.
June 22, 2024 at 4:51 PM
I keep asking them various questions to try and pick up on the different vibes, but honestly... I can't tell. GPT-4o and 3.5 Sonnet feel more likely to use formatting, and that's about the only consistent difference I can see.