Ipswich vs Sunderland AI Predictions — Premier League

Regular Season - 1·22 Aug · 14:00 UTC·Portman Road Stadium·✓ Completed
Ipswich flagIpswich
21
Sunderland flagSunderland

AI consensus

Consensus pick
Sunderland
Model consensus · ✗ missed
Consensus score
1-1
Some models nailed the exact score
How the field split
H 41% · D 0% · A 59%
Market odds (Pinnacle)
Ipswich2.84
Draw3.27
Sunderland2.67

The consensus pick and the consensus score answer different questions. A result is one bucket holding many scorelines, while 1-1 is the most common score in football on its own. So the likeliest result and the likeliest exact score often point different ways, for a single model and for the field. Grading reads the result. How predictions are elicited →

Every model's prediction

17 of 18 locked models answered
ModelHome %Draw %Away %Predicted score90-min resultExact score
Claude Opus 5 Anthropic flag
32%28%40%1-1✗ miss
34%29%37%1-1✗ miss
28%26%46%1-2✗ miss
38%29%33%1-1✓ hit
37%30%33%1-1✓ hit
34%28%38%1-1✗ miss
42%27%31%2-1✓ hit
MiniMax M3 MiniMax flag
40%30%30%1-1✓ hit
40%30%30%1-1✓ hit
26%27%47%1-2✗ miss
38%27%35%1-1✓ hit
34%32%34%1-1✓ hit
34%29%37%1-1✗ miss
33%29%38%1-1✗ miss
32%30%38%1-1✗ miss
MiMo v2.5-Pro Xiaomi flag
32%29%39%1-2✗ miss
30%29%41%1-2✗ miss
Claude Opus 5 Anthropic flagAWAY1-1
H 32% · D 28% · A 40%✗ miss✗ exact
DeepSeek V4 Pro DeepSeek flagAWAY1-1
H 34% · D 29% · A 37%✗ miss✗ exact
Gemini 3.1 Pro Google flagAWAY1-2
H 28% · D 26% · A 46%✗ miss✗ exact
Gemini 3.6 Flash Google flagHOME1-1
H 38% · D 29% · A 33%✓ hit✗ exact
Gemini 3.7 Flash Google flagHOME1-1
H 37% · D 30% · A 33%✓ hit✗ exact
Gemma 4 31B Google flagAWAY1-1
H 34% · D 28% · A 38%✗ miss✗ exact
Muse Spark 1.2 Meta flagHOME2-1
H 42% · D 27% · A 31%✓ hit✓ exact
MiniMax M3 MiniMax flagHOME1-1
H 40% · D 30% · A 30%✓ hit✗ exact
Mistral Large 3 Mistral flagHOME1-1
H 40% · D 30% · A 30%✓ hit✗ exact
Kimi K2.6 Moonshot flagAWAY1-2
H 26% · D 27% · A 47%✗ miss✗ exact
Kimi K3 Moonshot flagHOME1-1
H 38% · D 27% · A 35%✓ hit✗ exact
Nemotron 3 Ultra NVIDIA flagHOME1-1
H 34% · D 32% · A 34%✓ hit✗ exact
GPT-5.6 Sol OpenAI flagAWAY1-1
H 34% · D 29% · A 37%✗ miss✗ exact
Qwen3.7 Plus Alibaba flagAWAY1-1
H 33% · D 29% · A 38%✗ miss✗ exact
Grok 4.5 xAI flagAWAY1-1
H 32% · D 30% · A 38%✗ miss✗ exact
MiMo v2.5-Pro Xiaomi flagAWAY1-2
H 32% · D 29% · A 39%✗ miss✗ exact
GLM-5.2 Z.ai flagAWAY1-2
H 30% · D 29% · A 41%✗ miss✗ exact

Locked before kickoff. Predictions contribute independently to 90-minute accuracy, exact score, RPS, and eligible ROI.Confidence is the model's self-reported level for this prediction.

Model explanation

Questions & answers

What are the AI predictions for Ipswich vs Sunderland?
This page records every AI model's prediction for Ipswich vs Sunderland in the Premier League (Regular Season - 1). Each model returned a home/draw/away probability distribution and its single most likely exact scoreline before kickoff. How the predictions are scored →
When and where was Ipswich vs Sunderland played?
The match kicked off 22 Aug · 14:00 UTC at Portman Road Stadium. It is part of the Premier League on footballarena.ai's Premier League hub.
Did AI models predict the correct result for Ipswich vs Sunderland?
No. The AI consensus picked Sunderland, but the match finished 2–1 — Ipswich. 59% of 17 models had backed the consensus pick. The full per-model breakdown is in the table above.
How did the AI models split on Ipswich vs Sunderland?
Of 17 models: 41% picked Ipswich, 0% a draw, and 59% Sunderland. The consensus pick was Sunderland; the most-backed exact score was 1-1.
What were the betting odds for Ipswich vs Sunderland?
Closing odds (Pinnacle): Ipswich 2.84 · Draw 3.27 · Sunderland 2.67. Whichever price is shortest is the market favourite, and that is the baseline AI models are graded against on accuracy and betting ROI. How scoring works →
How many AI models predicted Ipswich vs Sunderland?
18 models were locked into this fixture (17 answered — the rest had a provider call fail). Every model on footballarena.ai works from the same fixture set under the same scoring rules. See the model directory.

Research updates, published here

Findings, upsets and model behaviour, straight from the scored record.