Houston 26–28 Texas Tech: AI predicted 23–32 ✓

Where the panel agrees — and it's unanimous

All five models back a Texas Tech win, and even more strikingly, all five independently converge on the same core insight: the -13.5 spread prices Texas Tech's 2025 reputation, not the 2026 September evidence. The single most persuasive datapoint, surfaced independently by every expert, is the common-opponent comparison with Oregon State: Houston beat the Beavers 33-20 at home (533 total yards, 228 rushing, only 26 rushing yards allowed), while Texas Tech won 35-24 in Corvallis while being outgained 472-352, allowing 417 passing yards, and surrendering four sacks of Will Hammond.

Resolving the disagreement: how big is the Tech edge?

The experts differ mainly on margin, and that split maps onto how much weight each gives to 2025 pedigree versus current tape. GLM-5.3 (38-24) and Kimi K3 (34-21) lean on Texas Tech's defensive front depth and the defending-champion roster construction, projecting a two-score pullaway. Grok 4.6, Claude Opus 5, and GPT-5.6 (27-23, 31-24, 31-24) counter with the more concrete current-season concerns: Tech's pass defense has been genuinely leaky (even Abilene Christian's receiver caught eight balls), depth corner Amier Boyd is out, edge rusher Trey White is questionable, and Hammond — a first-year full-time starter coming off an ACL recovery — is managing a run-first, low-variance offense that threw for just 189 yards at Oregon State. As Claude Opus 5 noted, that offensive profile wins games but rarely blows them open.

The synthesis favors the tighter read for three reasons. First, the injury and quarterback-form evidence is specific and verifiable, while the "Tech's front will suffocate them" case is more abstract — Houston's Conner Weigman has thrown for 300+ yards against the same defense Tech struggled with, and Grok 4.6's point that Houston's offensive line and Weigman's mobility should sustain drives is well-supported. Second, Houston has multiple credible paths to staying within two touchdowns: sustained rushing success, Weigman exploiting a thinned secondary, and a defense that has already proven it can erase an opponent's ground game. Third, the favourite-lean bias in these models is a known failure mode — when they cluster on a big-favourite call, the true gap is usually smaller.