25
Meta's new Llama 4 is actually good at coding, not just writing poems
I ran the same Python script through Llama 3.1 and GPT-4o last week and Llama 4 finished it in 12 seconds with zero bugs while GPT-4o gave me three broken imports, has anyone else noticed it handling APIs better?
2 comments
Log in to join the discussion
Log In2 Comments
angelac631mo ago
Gotta point out that GPT-4o isn't a single model, it's a whole system with different versions.
3
the_max1mo ago
Right, that's exactly what I've been saying. The "gpt-4o" label hides a lot of messy stuff. I've noticed the same thing - sometimes it feels like a totally different model depending on what mode it's in. One minute it's super fast and helpful, the next it's taking forever and giving me weird answers. It's like they're running a few different brains under the same name. Makes it hard to get a consistent feel for what it can actually do.
1