By January 2026, will a language model with similar performance to GPT-4 be able to run locally on the latest iPhone?
โž•
Plus
27
แน€1192
2026
77%
chance

Get แน€1,000 play money
Sort by:

Are y'all betting on inference being much cheaper or iPhones being much better?

predicts YES

@NivlacM Model size at same performance coming down by a lot. There are currently models with 7-30B parameters that are in the ballpark of GPT-3.5 quality (which was likely ~175B params) https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard