Which AI is best at
predicting sports?
Claude, GPT-5, Gemini, Grok and a dozen more call the same match from the same facts — then the final score grades every one, in public.
No tips, no cherry-picking — every win and loss in the open, so your next call is better informed.
- Picks graded
- 15,998
- Best AI right now
- GPT-4o Mini 64%
- Hindsight edits
- Zero
Free forever, no card. Not a sportsbook — we don't take bets or set odds, we just score the AIs against the final result.
🏆 Beat the AIs on the World Cup Pick exact scores · free entry · winners get a year of ProHow the AIs did
Wed, Jul 22's scorecard
TC TEAM COOP vs TS TEAM SPOON
· Women's National Basketball AssociationYour call before the AIs reveal theirs — who wins?
You picked . 🤝 AIs are with you. 🤖 But AIs took — bold call.
Save your call, follow how the AIs do, and play the prize game — free.
Lock my pick — sign up freeAI vs AI Top Picks
The top three matches the arena is calling right now.
Today's arena
Every AI's call on each match. Click for the full breakdown.
TEAM COOP vs TEAM SPOON
TS6/7 lean TEAM COOP
-
Gemini 2.5 Flash
TEAM COOP
55%
-
Grok 4 Fast
TEAM COOP
62%
-
Claude Haiku 4.5 TEAM COOP 58%
-
Gemini 2.5 Flash-Lite
TEAM COOP
60%
-
DeepSeek V3
TEAM COOP
62%
-
GPT-4o Mini
TEAM COOP
75%
-
GPT-5 Mini
Team Spoon
58%
Tampa Bay Rays vs Cleveland Guardians
5/7 lean Tampa Bay Rays
-
Grok 4 Fast
Tampa Bay Rays
54%
-
DeepSeek V3
Cleveland Guardians
55%
-
Claude Haiku 4.5 Tampa Bay Rays 58%
-
Gemini 2.5 Flash
Tampa Bay Rays
53%
-
GPT-5 Mini
Tampa Bay Rays
58%
-
Gemini 2.5 Flash-Lite
Tampa Bay Rays
58%
-
GPT-4o Mini
Cleveland Guardians
55%
Boston Red Sox vs Toronto Blue Jays
6/7 lean Boston Red Sox
-
Claude Haiku 4.5 Boston Red Sox 62%
-
GPT-4o Mini
Toronto Blue Jays
60%
-
Grok 4 Fast
Boston Red Sox
58%
-
Gemini 2.5 Flash-Lite
Boston Red Sox
55%
-
Gemini 2.5 Flash
Boston Red Sox
58%
-
DeepSeek V3
Boston Red Sox
60%
-
GPT-5 Mini
Boston Red Sox
62%
Washington Nationals vs Arizona Diamondbacks
4/7 lean Washington Nationals
-
Claude Haiku 4.5 Washington Nationals 55%
-
GPT-4o Mini
Arizona Diamondbacks
55%
-
Grok 4 Fast
Washington Nationals
53%
-
Gemini 2.5 Flash
Washington Nationals
52%
-
Gemini 2.5 Flash-Lite
Arizona Diamondbacks
55%
-
DeepSeek V3
Washington Nationals
55%
-
GPT-5 Mini
Arizona Diamondbacks
55%
Pittsburgh Pirates vs Chicago Cubs
5/7 lean Pittsburgh Pirates
-
Gemini 2.5 Flash
Pittsburgh Pirates
58%
-
Claude Haiku 4.5 Pittsburgh Pirates 55%
-
DeepSeek V3
Pittsburgh Pirates
55%
-
GPT-5 Mini
Pittsburgh Pirates
57%
-
Grok 4 Fast
Pirates
53%
-
GPT-4o Mini
Pittsburgh Pirates
55%
-
Gemini 2.5 Flash-Lite
Chicago Cubs
55%
Baltimore Orioles vs Atlanta Braves
4/7 lean Atlanta Braves
-
Grok 4 Fast
Baltimore Orioles
53%
-
DeepSeek V3
Baltimore Orioles
55%
-
Claude Haiku 4.5 Atlanta Braves 58%
-
Gemini 2.5 Flash
Atlanta Braves
62%
-
GPT-4o Mini
Atlanta Braves
57%
-
Gemini 2.5 Flash-Lite
Atlanta Braves
65%
-
GPT-5 Mini
Baltimore Orioles
62%
Miami Marlins vs San Diego Padres
5/7 lean Miami Marlins
-
Gemini 2.5 Flash
Miami Marlins
65%
-
DeepSeek V3
Miami Marlins
65%
-
Grok 4 Fast
Miami Marlins
58%
-
Claude Haiku 4.5 Miami Marlins 58%
-
GPT-4o Mini
San Diego Padres
55%
-
Gemini 2.5 Flash-Lite
San Diego Padres
55%
-
GPT-5 Mini
Miami Marlins
54%
New York Mets vs Los Angeles Dodgers
7/7 lean Los Angeles Dodgers
-
Gemini 2.5 Flash-Lite
Los Angeles Dodgers
55%
-
Claude Haiku 4.5 Los Angeles Dodgers 58%
-
GPT-4o Mini
Los Angeles Dodgers
75%
-
Grok 4 Fast
Los Angeles Dodgers
58%
-
Gemini 2.5 Flash
Los Angeles Dodgers
58%
-
DeepSeek V3
Los Angeles Dodgers
58%
-
GPT-5 Mini
Los Angeles Dodgers
68%
Milwaukee Brewers vs Colorado Rockies
7/7 lean Milwaukee Brewers
-
Claude Haiku 4.5 Milwaukee Brewers 62%
-
Grok 4 Fast
Milwaukee Brewers
67%
-
Gemini 2.5 Flash-Lite
Milwaukee Brewers
65%
-
GPT-4o Mini
Milwaukee Brewers
75%
-
DeepSeek V3
Milwaukee Brewers
65%
-
Gemini 2.5 Flash
Milwaukee Brewers
68%
-
GPT-5 Mini
Milwaukee Brewers
72%
Chicago White Sox vs Houston Astros
7/7 lean Houston Astros
-
DeepSeek V3
Houston Astros
58%
-
Grok 4 Fast
Houston Astros
62%
-
Gemini 2.5 Flash
Houston Astros
58%
-
Claude Haiku 4.5 Houston Astros 62%
-
Gemini 2.5 Flash-Lite
Houston Astros
65%
-
GPT-4o Mini
Houston Astros
55%
-
GPT-5 Mini
Houston Astros
56%
Minnesota Twins vs Athletics
6/7 lean Minnesota Twins
-
Grok 4 Fast
Minnesota Twins
62%
-
DeepSeek V3
Minnesota Twins
62%
-
Claude Haiku 4.5 Minnesota Twins 72%
-
Gemini 2.5 Flash
Minnesota Twins
70%
-
GPT-4o Mini
Athletics
55%
-
Gemini 2.5 Flash-Lite
Minnesota Twins
60%
-
GPT-5 Mini
Minnesota Twins
65%
St. Louis Cardinals vs Cincinnati Reds
6/7 lean St. Louis Cardinals
-
Grok 4 Fast
St. Louis Cardinals
54%
-
DeepSeek V3
St. Louis Cardinals
58%
-
GPT-4o Mini
St. Louis Cardinals
55%
-
Gemini 2.5 Flash
St. Louis Cardinals
55%
-
Claude Haiku 4.5 St. Louis Cardinals 58%
-
Gemini 2.5 Flash-Lite
Cincinnati Reds
55%
-
GPT-5 Mini
St. Louis Cardinals
58%
How it works
Same brief. Independent calls. Public scoreboard.
-
Step 01
Same brief, every model
For each match we render one prompt — teams, injuries, recent form, market line — and send it byte-for-byte to every AI in the arena. A prompt hash is stored on every prediction as the integrity proof.
-
Step 02
Independent picks
Each model returns its pick, confidence, probability distribution and a written rationale. No cross-talk, no editorial layer. The full response is captured at call time so we can audit later.
-
Step 03
Graded by reality
Once the result lands we settle every pick automatically: win rate, units, and Brier-score calibration. Picks are permanent and public — no hindsight edits, no hiding misses.
Recently settled
Receipts. The leaderboard isn't theatrical.
AI streaks
Last 10 settled picks per model — who's hot, who's cold.
🔒 Pro lineup
Flagship AIs in your corner
The frontier models from Anthropic, Google, OpenAI and xAI — all calling the same matches on the same brief. Pro unlocks their picks + lets you trigger any flagship on any match on demand.
-
Claude Opus 4.7
ProFlagship reasoning
-
Gemini 2.5 Pro
ProDeepest context
-
GPT-5
ProFastest analysis
Pro · from $9/mo
Free shows the picks.
Pro shows the edge.
Get pinged when AI consensus disagrees with the market, see why each model picked what it picked, and track your own ROI side-by-side with every frontier AI.
Every pick is locked before kickoff and graded in public — losses included. No cherry-picking, no hindsight edits — unlike any tipster, the full record is on the table before you pay.
Model-vs-market alerts
Push notification the second AI consensus beats market price by your chosen edge.
AI reasoning unlocked
See every model's chain-of-thought, confidence, and the sources it cited.
ROI tracker
Your ROI and CLV — side-by-side with every frontier AI over the same period.
From the blog
Methodology, weekly recaps, AI vs AI analysis.
-
AI PredictionsWhy Sports Tipsters Profit When You Lose — and What Actually Works
Around 95% of sports bettors lose money. Many tipsters are bookmaker affiliates who earn 10–40% of your losses — so you lose twice. The math, the sources, and a fully transparent alternative.
Jun 23, 2026 7 min read -
AI PredictionsEcuador 0-0 Curaçao: The AI Panel Went 7-for-7 on Ecuador — And All Seven Were Wrong
Seven frontier AIs unanimously tipped Ecuador, several with 85%+ confidence. The match ended 0-0 — a clean sweep of misses and a textbook lesson in consensus overconfidence.
Jun 21, 2026 6 min read -
AI PredictionsSpain 4-0 Saudi Arabia: The AI Panel Went 15-for-15 on the Winner and Still Missed the Scoreline
Every model on the ModelFights panel backed Spain, and Spain duly delivered a 4-0 rout. But the unanimous winner call hid a clean sweep of misses on the exact scoreline.
Jun 21, 2026 5 min read
FAQ
Questions, answered.
-
What is ModelFights?
A live scoreboard for AI sports prediction. Frontier models — Claude, GPT, Gemini, Grok, DeepSeek and more — call every match from the same brief. Picks lock at kickoff. Public, no hindsight. -
How do AIs make the picks?
Every model receives an identical structured brief (teams, recent form, line-up news, market odds, weather). We publish the prompt + SHA-256 hash on every match page so anyone can verify nothing was tampered with after results landed. -
Are picks really locked before kickoff?
Yes. Each prediction is timestamped before kickoff and immutable in the database. The grading layer reads results from independent feeds and settles automatically — no human can edit a pick after the fact. -
What's the difference between free and Pro?
Free shows you every regular-tier model's pick on every match. Pro ($9/mo Follower · $29/mo Insider · $99/mo Analyst) unlocks flagship-model picks (Claude Opus, GPT-5, Gemini Pro), on-demand audits where you trigger any model on any match, and pre-kickoff model-vs-market alerts. Every Pro plan starts with a 7-day free trial — cancel any time before it ends and you pay nothing. -
How does the "Beat the AIs" competition work?
Free to enter. Pick the exact score on every World Cup match. 5 points for an exact score, 3 for correct winner + goal difference, 1 for correct winner only. Top scorer at tournament end wins a lifetime free plan; every qualified entrant who outscores the AI lineup also gets a year of Pro. -
Where do the odds come from?
We pull live odds from The Odds API across 15+ bookmakers (Pinnacle, Stake, bet365, DraftKings, FanDuel and others). Edge percentages compare each AI's probability to the best available bookmaker line. -
Can I track my own picks?
Yes. Sign up free, save matches with the ★ button, place exact-score picks on World Cup matches via the competition, and check your public profile at /u/{your-username} for streaks + cumulative points. -
Do you offer an API?
API access ships with the Analyst plan ($99/mo). Lower tiers get JSON exports of their own pick history. Reach out via support if you need higher-volume access for research.
🔥 DeepSeek V3 is on a 4-win streak
Free account. Get notified the moment they lock their next pick.