Skip to content
Premier League ·

Manchester United vs Manchester City prediction: 5 of 5 AI models back Manchester City

Updated · ScoreGPT multi-model panel
Manchester United
1–2AI consensus
Manchester City

Download the free app

Every model's pick, full analysis — free to start.

What does the AI predict for Manchester United vs Manchester City?

As of 12 September 2026, for Manchester United vs Manchester City (Premier League), ScoreGPT's AI panel reached a 1–2 consensus scoreline, with 5 of 5 models picking Manchester City. Key factors: United missing de Ligt, Ugarte — gutted defensive spine; Haaland mismatch vs Martinez–Maguire pairing. Every pick is graded publicly after full-time — information only, not betting advice.

Every pick is graded after full time — see how our grading works and the live AI model leaderboard.

What you get in the app

Every model's reasoning — not just the score

See all model picks, the full breakdown, and how each one performed — free.

Every AI model’s scoreline pick and its confidence for the same match in the ScoreGPT app
ScoreGPT model rankings screen with rolling outcome accuracy for every AI model

See the full analysis — free

The AI verdict

Manchester City are the lean, but only narrowly. City's superior form and United's injury-hit defensive spine tilt the derby, yet Old Trafford precedent and derby compression keep United live.

  • City arrive top of the table with three wins from three, while United sit 11th with four points.
  • United are missing De Ligt and Ugarte, gutting their defensive spine; Baleba is doubtful.
  • City have two extra days' rest after playing Tuesday, while United played Thursday.
  • United have scored fourteen and conceded nine in their last five home league games.
  • Carrick has won nine of his ten Premier League home matches, including a win over City in January.

AI analysis generated 2026-09-12

Why ScoreGPT's AI panel leans this way — Manchester United vs Manchester City

  • United missing de Ligt, Ugarte — gutted defensive spine
  • Haaland mismatch vs Martinez–Maguire pairing
  • City two extra days' rest before derby
  • United open at home: 14 scored, 9 conceded in last five

The read: a tilted derby, not a mismatch

All five models converge on a narrow Manchester City win, and crucially, none of them arrived there casually — every expert held the away win below the 50% threshold despite City's perfect start. That restraint is the correct calibration for a fixture where the raw form gap is real but the context keeps United live. City arrive top of the table with three wins from three (7-2 aggregate) and a composed 2-0 win in Porto on Tuesday; United sit 11th with four points, a home loss to promoted Hull, and a 2-2 at Everton where they led twice and were pegged back late. Yet Carrick has won nine of his ten Premier League home matches, including a 2-0 win over City in this exact fixture in January, and City have won only two of the last six derbies. That tension — structural City superiority against genuine Old Trafford precedent — is what keeps the draw and home win each at a meaningful 27%.

Where the experts resolved the same question the same way

The panel independently identified the same decisive levers, which strengthens confidence in the read even if it doesn't justify inflating it:

  • United's gutted defensive spine. De Ligt (back) and Ugarte (knee) are out, Baleba doubtful. United have conceded six in three league games with those players partly available. The Martinez–Maguire pairing lacks the pace and aerial dominance to handle Haaland's back-shoulder runs, and the projected Mainoo–Tielemans midfield has no natural ball-winner against a City side built to attack the zone between midfield and defence.
  • The rest asymmetry. United played Thursday (4-0 vs Sabah); City played Tuesday in Porto. In a high-intensity derby with a 15:30 Sunday kickoff, those extra 48 hours typically surface in the final half hour — precisely when a United team missing its ball-winners is most exposed to transitions.
  • United's open home profile cuts both ways. Fourteen scored, nine conceded in their last five home league games. They are entertaining and beatable — which is why every model's modal scoreline (1-2) features goals at both ends rather than a City clean sheet.

The model panel: Manchester United vs Manchester City

Premier League · 13 September 2026 · scorelines read Manchester United–Manchester City

GPT-5.6
1–2
Claude Opus 5
1–2
Grok 4.6
Leans Manchester City
🔒 exact score in app
Kimi K3
Leans Manchester City
🔒 exact score in app
GLM-5.3
Leans Manchester City
🔒 exact score in app

Unlock all 5 exact model picks free in the app →

Frequently asked

Who is more likely to win, Manchester United or Manchester City?

Manchester City are the narrow lean. Their perfect start and United's missing defensive spine give them the edge, but the gap is compressed by United's strong home record under Carrick and the unpredictable nature of derbies.

What is the predicted score for Manchester United vs Manchester City?

The averaged AI scoreline is Manchester United 1–2 Manchester City.

How does ScoreGPT make Premier League predictions?

Multiple AI models independently analyze form, squads and match context for every covered Premier League fixture. ScoreGPT compares their picks and grades every prediction publicly after the final whistle.

What do the AI models predict for Manchester United vs Manchester City?

All five models converge on a narrow Manchester City win, but none exceed the fifty percent threshold. They see an open, transitional game where United can score, yet City's structural superiority and extra rest make them the likeliest single outcome.

What are the key factors for Manchester United vs Manchester City?

United are without De Ligt and Ugarte, leaving a slow centre-back pairing against Haaland. City have two extra days' rest. United are open at home, scoring and conceding freely. Carrick's strong home record and a recent derby win keep United dangerous.

More Premier League predictions

Recent Premier League results — graded

AI predictions are for information and entertainment only — not betting advice. 18+. Please play responsibly.