Osasuna 0–0 Levante: AI predicted 2–1 ✗

Unanimous on direction, divided on shape

All five models independently land on a home win, and that convergence carries weight — but it also maps neatly onto a market that has already priced Osasuna at 1.80 (~52-53% implied). The more interesting disagreement is how the win arrives: GPT-5.6 and Grok 4.6 see a compressed 1-0 built on Ramis's defensive identity and Levante's shocking opening-day sterility (zero shots on target at Espanyol, 0.28 xG), while Claude Opus 5, Kimi K3, and GLM-5.2 see a 2-1 driven by both teams' chronically leaky defenses and a pattern of high-scoring recent fixtures.

Resolving the goals question

The 1-0 case rests on three pillars: Ramis's Burgos teams were genuinely compact and set-piece reliant; Levante just produced one of the most anemic attacking performances of the opening weekend; and first competitive games under conservative new managers tend toward cagey, low-event affairs. These are real arguments, and Grok — the strongest recent performer at 55% result accuracy — is their champion.

But the 2-1 case has the deeper data behind it. Osasuna conceded in 100% of their last six matches (11 goals total), and a 92-day competitive layoff under a new coach is precisely the condition that produces defensive disorganization — the opposite of the rust-proofing GPT implies. Levante's 0-3 at Espanyol is alarming but distorted: they were missing chief creator Carlos Álvarez, had unresolved registrations, and Roger Brugué was suspended. Brugué returns here, and additional registrations may clear. Levante scored three at Osasuna in May and three at Celta — the attacking capacity exists even without departed top scorer Carlos Espí. Meanwhile, 83% of each team's recent matches went over 2.5 goals, and the last head-to-head finished 3-2.