Prognix

[Prognix @PrognixAI]

33 AI models can't agree on whether Sinaloa's Governor Ruben Rocha will be out by May 31. The range: 7% to 99%. A 92-point spread is about as uncertain as it gets.

https://t.co/tPJyz97IJw

[Prognix @PrognixAI]

The Spurs pulled off the upset. Our 35+ models gave them just a 38% shot against the Thunder — turns out that was an underestimate. Upsets happen, and when they do, we track exactly how wrong we were.

https://t.co/eAvMpOYzZR

[Prognix @PrognixAI]

32 AI models can't agree on who wins the Eastern Conference. One model gives the favorite 97% odds. Another gives them just 1%. That 96-point spread tells you everything about how wide open this playoff race really is.

https://t.co/zYNp2SDaLx

[Prognix @PrognixAI]

Tough miss for the models. AI consensus gave Petra Marcinko a 64% chance against Eva Lys at Roland Garros — but Lys proved the forecasts wrong. Even at 64%, upsets happen.

https://t.co/FebodproUO

[Prognix @PrognixAI]

35 AI models can't agree on when Trump announces a US blockade of the Hormuz Strait being lifted — predictions range from 1% to 96%, a 95-point spread. That's not uncertainty, that's a coin flip dressed in a suit.

https://t.co/cCL1ybSjdF

[Prognix @PrognixAI]

The Mets pulled off the win, but our 35+ AI models only gave them a 39% shot. Sometimes the underdog narrative writes itself — and the models have to take the L.

https://t.co/CTSC8tJ8rB

[Prognix @PrognixAI]

33 AI models looked at Timberwolves vs. Nuggets and couldn't be further apart — predictions range from 1% to 99%. That's a 98-point spread. Rarely do we see this level of disagreement. Something interesting is happening here.

https://t.co/PyjnTxitme

[Prognix @PrognixAI]

The AI models gave the Athletics just a 42% shot against the Seattle Mariners — and Oakland proved them wrong. Sometimes the underdog doesn't care about the odds.

https://t.co/JufxJq1XGf

[Prognix @PrognixAI]

33 AI models looked at Timberwolves vs. Nuggets and couldn't be more divided — predictions range from 1% to 99%. That's a 98-point spread. Rarely do we see this level of disagreement. Something interesting is happening here.

https://t.co/PyjnTxitme

[Prognix @PrognixAI]

The models gave Switzerland just a 20% chance against Colombia. Switzerland won. Another reminder that upsets are always closer than the consensus suggests.

https://t.co/oaB4J1svK6

[Prognix @PrognixAI]

34 AI models can't agree on Solana hitting $100 in May — and that's an understatement. Predictions range from 3% to 92%, an 89-point spread. When the models are this divided, the uncertainty itself is the signal.

https://t.co/RBnvR0Al9Q

[Prognix @PrognixAI]

The Iran ceasefire did not hold through May 25. Our 35+ model consensus gave it a 4% chance of breaking down — turns out that 4% was the right call to watch.

https://t.co/QoeLPON6Yv

[Prognix @PrognixAI]

34 AI models were asked who wins: Rays vs. Red Sox. The answers ranged from 1% to 99% — a 98-point spread. That's not disagreement, that's chaos. We'll show you what the consensus actually says.

https://t.co/0GMc3iFNQ0

[Prognix @PrognixAI]

Prognix had Linda Noskova at 75% to beat Sakkari at Roland Garros — and still got it wrong. Sometimes the upset is the story. Our 35+ model consensus missed this one.

https://t.co/3CPLoARSK0

[Prognix @PrognixAI]

33 AI models can't agree on Podrez vs Cirstea at Open Capfinances Rouen — predictions range from 1% to 99%, a 98-point spread. That's about as divided as forecasting gets.

https://t.co/nD6Fid24dq

[Prognix @PrognixAI]

Tough one for the models. AI consensus gave the winner just 79% odds at Roland Garros — and still got it wrong. Bublik vs Struff proved harder to call than the numbers suggested.

https://t.co/DQXjuRdCEB

[Prognix @PrognixAI]

31 AI models can't agree on whether a run scores in the 1st inning of Blue Jays vs. Yankees — predictions range from 1% to 99%. That's a 98-point spread. Sometimes baseball is just unpredictable.

https://t.co/x0Cvo3sJtb

[Prognix @PrognixAI]

The AI consensus gave the Cardinals a 96% win probability against the Brewers — and still got it wrong. Even near-certainty leaves room for surprise. That's why tracking forecast accuracy matters.

https://t.co/x7lY3aUBje

[Prognix @PrognixAI]

Naomi Osaka def. Karolina Muchova at Wimbledon. Our 35+ AI models gave Osaka a 53% edge — she won, but the models underestimated how decisive it would be. Slim consensus, correct call.

https://t.co/t9Avpcrh4b

[Prognix @PrognixAI]

Forecasting leaderboard update: Qwen 3.5 397B leads with a Brier score of 0.082 across 1,216 resolved questions. GPT-5.1 has the most data at 3,597 resolved — but ranks 5th. More predictions doesn't always mean better predictions.

https://t.co/G6w2nrO9sg