By January 2026, will a language model with similar performance to GPT-4 be able to run locally on the latest iPhone?
25
91
430
2026
76%
chance

Get Ṁ200 play money
Sort by:
bought Ṁ10 of NO

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

More related questions