FRFrance
W
LWWDW
MOMorocco
L
WDWWW
W3
Thu 9 Jul, 8:00 pm UTC
2 - 0
Full time
10/120 AIs back France · 64%
This match is over — see which AIs called it right.7 AI models predict every match, graded in public. Make them call today's upcoming football — before the matches start.Today's football
Value readAI consensus vs the market
On France
AI consensus64.0%
Market implied58.8%
+5.2%
AI edge

The AIs rate France 5.2 points more likely than the market's price implies — a positive edge.

The AI verdict
15 frontier models · 120 calls
ModelMatch winnerOver / Under
ConsensusFrance10/15 modelsUnder 2.56/15 models
Grok 4.3FLAGSHIPxai · France72%Under58%
Grok 4 FastFLAGSHIPxai · 48% · 4,469 graded · France62%Under55%
Claude Sonnet 4.6FLAGSHIPanthropic · France62%Under 2.570%
Claude Opus 4.7FLAGSHIPanthropic · France60%Under 2.556%
Claude Opus 4.6FLAGSHIPanthropic · France60%Under 2.558%
Claude Opus 4.8FLAGSHIPanthropic · France60%Under57%
GPT-5FLAGSHIPopenai · France59%Under 2.566%
o4-miniopenai · France75%Under 2.560%
DeepSeek V3deepseek · 46% · 5,379 graded · France70%Over60%
Gemini 2.5 Flash-Litegoogle · 46% · 3,706 graded · France65%Over55%
Gemini 2.5 Flashgoogle · 48% · 4,265 graded · France65%Under 2.560%
GPT-4o Miniopenai · 54% · 2,296 graded · France65%Under55%
Gemini 2.5 Progoogle · France62%Under58%
GPT-5 Miniopenai · 53% · 1,424 graded · France62%Over56%
Gemini 3.1 Progoogle · France60%Under55%
Form — last 5

France

WDWWL

Morocco

WWWDW
This match is over

The result is in, so the AIs can't be called on it — every call locks before the match to stay honest. Put them to work on today's upcoming games instead.

Market implied probabilityPolymarket · live
France
60%1.70
15
Draw
25%4.00
Morocco
15%6.81
Recent tradesPolymarket · live
Loading trades…
1X2Over/UnderBoth teams to scoreDouble chance

Moneyline (1X2) — best available price, with the market's implied probability.

OutcomeDecimalImpliedAI edge
France1.7058.8%+5.2%
Draw4.0025%—
Morocco6.8114.7%—

AI edge = the model consensus' probability minus the market's implied probability on the same outcome. Positive edge means the AIs rate it more likely than the book.

System instruction

You are a sports forecasting model. You are given identical structured facts about one match. Make an honest, calibrated prediction. You may not browse; use only the facts provided. Do not hedge — commit to one pick and a probability.

User brief (JSON) — sent byte-for-byte to every model

{
  "match": {
    "home": "France",
    "away": "Morocco",
    "sport": "football",
    "league": "World Cup",
    "kickoff_utc": "2026-07-09T20:00:00.000Z",
    "venue": "Gillette Stadium"
  },
  "market": {
    "type": "moneyline",
    "options": [
      "home",
      "draw",
      "away"
    ]
  },
  "bookmaker_odds": {
    "home": "1.7",
    "draw": "4",
    "away": "6.81"
  },
  "instructions": "Return your pick, a 0–100 confidence, a probability for each outcome, and one paragraph of reasoning."
}

A SHA-256 hash of this exact brief is stored on every prediction before kickoff. Nothing here changes after the result lands — hash it yourself to verify.

Every prediction here is made by a frontier AI model from the same structured brief, locked before kickoff with a timestamped integrity hash, and graded in public against the real result — losses included. No tips, no hindsight, no sportsbook money.