A short, speculative YouTube discussion by Matt Wolfe on whether and when Meta's Llama-3 might surpass OpenAI's GPT-4. The speakers consider hardware scaling (large H100 purchases), iterative upgrades to closed-source models (OpenAI/Anthropic/Gemini), and uncertainty about how the performance gap will evolve as newer models (GPT-5) emerge.
When can llama3 beat GPT4? That's the right question to ask. Maybe it's this year, Maybe. You know, I hear they're trying their best, but given that he's purchased 600,000 H1 hundreds, it's inevitable that he beats them. Right? Like it's just a matter of time now. Then you can say, okay, by the time he beats them, will Sam have a better model? Definitely. Like they've already had a year for GPT4 and they've been upgrading GPT4 through the course of the year, but they had a year to an even better model. So most likely there'll be a version of closed source either as OpenAI or anthropic or Gemini that will be better than the best llama at that point. But how long is that gap going to constantly be the case and how much would the gap be? Would the gap between GPT5 and llama 3 be even bigger than GPT4 and llama 2? It's not clear.
AI News: Dots, GPT-6.1 Sol, Sonnet 5.5, Gemini 4, and everything you need to know
