Utah Tech 7–63 BYU: AI predicted 7–51 ✓

Synthesis: A Mismatch Defined by Substitution Timing

All four models agree unanimously that BYU wins, and the disagreement is purely about margin — a gap that matters enormously for prediction but not at all for the outright result. The predictions cluster into two camps: a blowout camp (Grok 4.6 at 59-3, GPT-5.6 at 56-3) that sees BYU's starters running up a score approaching 60, and a controlled-win camp (Kimi K3 at 38-10, Claude Opus 5 at 52-10) that accounts for Kalani Sitake's well-documented tendency to empty the bench early in cupcake games.

Resolving the margin debate

The controlled-win camp has the stronger analytical case. Sitake's historical pattern is clear: last year's 69-0 demolition of Portland State is the exception, not the norm, and even in that game the scoring pace slowed dramatically once backups entered. BYU has a Big 12 conference slate beginning the following week, giving Sitake every incentive to rest Bear Bachmeier, LJ Martin, and key defensive starters by the third quarter. A run-heavy script once the lead is established — exactly what Kimi K3 and Claude describe — will chew clock and suppress the total below what a raw talent-gap analysis would suggest.

The blowout camp's case rests on Utah Tech's offensive futility — just 227 total yards and zero points against Montana State — and the likelihood that BYU's defense, which intercepted 17 passes in 2025, forces multiple turnovers that create short fields. That is a real factor, but it pushes the ceiling, not the median outcome. GPT-5.6 (55% 7-day accuracy, best in the group) and Grok 4.6 (51% 7-day, worst in the group) both lean into worst-case scenarios for Utah Tech without adequately weighting the pace suppression that bench-clearing creates.