Brad Wardell @draginol · 03 Sep 2026
I can flesh this out into an article if you guys want but 5 quick points on local AI and what's coming: 1. The smaller models have gotten drastically better in the past 90 days. I'd argue a 12B model, which can run on most hardware, is about as good as GPT 4 was for most uses.
1 345Views
35Likes
5Reposts
5Replies
0Quotes
7Bookmarks
Is that a lot?
1.80×vs this author's median749 views is typical
54Percentile for this authorof 11 recent posts
0.52×vs 10K–100K median2 578 views is typical
2.22%Reachviews ÷ followers
3.35%Engagement rateof viewers reacted