99% of users won't ever need anything beyond the lowest model
September 2, 2026
0:35
The AI model race looks like it's about raw intelligence. My co-host Nico's read on the show today is that it's actually a race to a finish line, after which most of it stops mattering.
His claim, direct: "It's like a tight race to finishing the SWE benchmark... once the SWE benchmark for all of them is 100%, and that benchmark doesn't mean anything anymore... 99% of users won't ever need to use anything beyond whatever the lowest model is at that point. Right?"
Whether or not that number holds, it's a genuinely different way to think about which model you actually need.
Where this came from
Value-First AI Daily - Sep 2, 2026Watch the full conversation this clip was pulled from.
The next step