TL;DR — All 6 AI models predict a Manchester City victory over Southampton in the FA Cup semi-final at Wembley. The consensus scoreline is 2-0. Three of five individual models call a clean sheet, and the ScoreGPT consensus flags BTTS-No as the standout value play at ~1.85 odds. Confidence is moderate (55-58%), held back by City's injury doubts (Rodri, Dias) against a Southampton side riding a remarkable 20-game unbeaten run.
Match Overview
Competition: FA Cup — Semi-Final
Date: Saturday, 25 April 2026
Kickoff: 16:15 UTC
Venue: Wembley Stadium, London
Premier League leaders Manchester City take on Championship promotion-chasers Southampton in a classic David-vs-Goliath semi-final at the national stadium. City have already won the Carabao Cup this season and are chasing a domestic treble, while Southampton — 4th in the Championship — are making their first Wembley FA Cup appearance since 2020/21.
What 6 AI Models Predict
| Model | Result | Predicted Score | Confidence | |---|---|---|:---:| | ScoreGPT (Consensus) | Home Win | 2–0 | 55% | | GPT-5.4 | Home Win | 2–0 | 58% | | Claude Opus 4.6 | Home Win | 2–0 | 55% | | Gemini 3.1 Pro | Home Win | 2–0 | 58% | | Grok 4.20 | Home Win | 2–1 | 28% | | Kimi K2.5 | Home Win | 3–1 | 62% |
Result agreement: 6/6 models predict a Manchester City win.
Score agreement: 4/6 models land on exactly 2-0.
Clean sheet consensus: 4/5 individual models predict Southampton fail to score (Kimi K2.5 is the outlier with a 3-1 prediction).
Consensus Analysis
The AI models are nearly unanimous on the outcome but diverge slightly on the margin of victory. Here's the core reasoning:
Why City wins: The quality gap between Premier League leaders and a Championship side is enormous, especially given the specific personnel losses Southampton are carrying. Flynn Downes (midfield anchor) and Ryan Manning (left-back and primary set-piece taker) are both suspended, stripping the Saints of their engine room and their best dead-ball outlet. Against a City side that has conceded just 1 goal in 5 matches and holds a staggering 21 consecutive FA Cup wins against non-Premier League sides, the task is monumental.
Why the score stays low: Four models (GPT-5.4, Claude Opus 4.6, Gemini 3.1 Pro, and the ScoreGPT consensus) all settle on a tidy 2-0. The reasoning is tactical: Guardiola's Wembley record in semi-finals tends toward controlled, conservative wins rather than blowouts (think 3-0 vs Sheffield United, 1-0 vs Brighton). City face a short 72-hour turnaround after a midweek win over Burnley, and Rodri (groin) and Ruben Dias (ankle) are doubtful — meaning Guardiola may opt for a more measured approach.
The contrarian view: Kimi K2.5 is the most bullish on City, predicting 3-1 with 62% confidence — the highest individual model confidence of the group. Kimi notes that City's firepower (Haaland with 24 goals this season, plus Semenyo and Cherki) could overwhelm a weakened Southampton backline, and that the Saints' 20-game unbeaten run was built mostly against Championship defences.
Grok 4.20 offers the lowest confidence at 28%, citing City's potential rotation and the risk of Southampton's momentum creating an upset — but still lands on a home win.
Value Detection
Value detected: BTTS-No at ~1.85
The ScoreGPT consensus identifies a meaningful gap between market pricing and AI probability on Both Teams to Score — No. The BTTS-No market sits at around 1.85 (implying ~54% probability), but our synthesis points to a true probability of 60-65%, creating an estimated edge of 6-10 percentage points.
Why the edge exists: The market is overweighting Southampton's 20-game scoring streak (built against Championship defences) and underweighting the specific impact of the Downes and Manning suspensions. City's clean sheet rate against lower-league FA Cup opposition sits around 55-60%, and their current defensive form — 1 goal conceded in 5 matches — is elite even by their standards.
Three independent models (GPT-5.4, Claude Opus 4.6, and the ScoreGPT consensus) all independently flagged BTTS-No as their primary bet opportunity, with Gemini 3.1 Pro offering a related play: Home Win & Under 3.5 goals at 1.95.
No significant value was detected on the outright home-win market (priced around 1.22), where AI win probabilities of 73-79% align relatively closely with market implied odds.
Frequently Asked Questions
Q: Who is predicted to win Manchester City vs Southampton?
All 6 AI models — GPT-5.4, Claude Opus 4.6, Gemini 3.1 Pro, Grok 4.20, Kimi K2.5, and the ScoreGPT consensus — predict a Manchester City victory. The consensus scoreline is 2-0, with four of five individual models forecasting a City clean sheet. Kimi K2.5 is the only model predicting both teams to score (3-1).
Q: What is the FA Cup semi-final history between these teams?
City have won their last three FA Cup semi-finals at Wembley, part of a wider record of 21 consecutive FA Cup wins against non-Premier League opposition (84 goals scored, 11 conceded). Southampton, by contrast, have lost 8 of 11 matches at Wembley — including both previous FA Cup semi-finals (to Chelsea in 2018 and Leicester in 2021). Their only Wembley win was the 2024 Championship playoff final.
Q: What are the key injuries and suspensions for this match?
Southampton are without Flynn Downes (suspended — violent conduct, 3-match ban) and Ryan Manning (suspended), their key midfielder and left-back respectively. This strips the Saints of their midfield engine and primary set-piece delivery. For City, Rodri (groin) is a serious doubt after missing the midweek Burnley game, Ruben Dias (ankle) is unlikely to return, and Josko Gvardiol (tibial fracture) is out for the season. Haaland's peculiar drought of 0 goal contributions in 14 semi-finals/finals for City (1,140 minutes, 25 shots) is also a statistical curiosity worth monitoring.
Disclaimer
Predictions by ScoreGPT using 6 independent AI models. For informational and entertainment purposes only. Download ScoreGPT free on iOS.