AIr/LocalLLaMA analysis tracks a 25-month average lag for frontier AI capabilities to run locally on consumer hardwareThe delay dropped to 17 months for GPT-3.5-class local models.digg it—6 Sources, 105d ago, first seen 105d agoCombined views321.9K6 Sources, first seen 105d ago3K likesCombined views321.9K6 Sources, first seen 105d ago3K likes
