April 13 recap: 20-10 combined for +9.18u. The best single day of the 2026 season. Plus the Tuesday board preview with 23 picks across a 15-game slate.
Twenty wins on ten losses. Plus 9.18 units on a Monday slate. That's the best combined day any of these four AIs have posted since the season started, and it wasn't even close. The combined record improved from -115.16u to -105.98u in a single night. If you've been waiting for a sign that this contest isn't just a slow bleed, Monday was it.
Claude ran the table. Nine wins against four losses, good for +5.10u, and the run line strategy finally paid dividends. Run line sweeps in Seattle (SEA ML + SEA -1.5), Baltimore (BAL ML + BAL -1.5), Philadelphia (PHI ML + PHI -1.5), and Texas (TEX ML + TEX -1.5) all cashed. That's four games where Claude doubled up on the winner at plus money. When those hit, the math gets beautiful in a hurry.
ChatGPT turned in a 7-3 day for +4.68u. The widest slate coverage on the board produced the most raw wins. The Mariners, Orioles, Phillies, Yankees, Guardians, TEX/OAK Over, and Dodgers all landed. The damage came from Atlanta (-156, 2u) and Boston (-168, 1.5u), both favorites that couldn't hold up. The Braves blew a lead, and the Red Sox got obliterated 6-13 at Minnesota.
Grok kept it clean at 2-1 for +0.50u. Three plays, two wins, one loss. Philly cashed, the TEX/OAK Over hit, and the Boston dog at +141 was the lone miss as the Twins ran the Red Sox off the field. Small but positive. That's all you can ask from a three-play card.
Gemini had the day's single biggest disaster. The 3u top play on Houston ML at -115 got torched as the Astros couldn't get anything going against Seattle. That's -3.45u on one bet. Gemini still went 2-2 overall, with Philly and the BOS/MIN F5 Over cashing, but the CHC/PHI Under took a loss too. Net: -1.10u. When your biggest conviction play is the day's worst loss, it stings.
The unanimous Phillies play was the day's safest bet. All four AIs backed Philadelphia, and Philly rolled Chicago 13-7. The rare 4-AI consensus went 1-0 on the season. You love to see it.
The one thing all four models agreed on beyond Philly? Boston was a good play at Minnesota. They were all wrong. The Red Sox got absolutely buried 6-13. Sometimes the whole room is wrong together.
| # | Model | Record | Units | Win% |
|---|---|---|---|---|
| 1 | Gemini | 40-38-4 | -15.66u | 51.3% |
| 2 | ChatGPT | 47-44-7 | -19.30u | 51.6% |
| 3 | Claude | 53-63-7 | -31.90u | 45.7% |
| 4 | Grok | 49-60-4 | -39.12u | 45.0% |
Combined: 189-205-22 | -105.98u
Gemini holds the lead at -15.66u despite yesterday's rough outing, but ChatGPT closed the gap to just 3.64u with that 7-3 day. Claude sits 16.24u back in third, but Monday's surge was the kind of day that reminds you this thing isn't over. Grok's conservative approach keeps the damage manageable but the hole is deep at -39.12u.
Fifteen games. Twenty-three total picks across the four models. And the consensus is as tight as we've seen all season.
Three AIs love Pittsburgh. Grok (2u at -186), ChatGPT (2u at -186), and Claude (2u at -150) all back the Pirates against Washington. The edge is obvious: Mitch Keller's 1.00 ERA against Miles Mikolas and his 12.41 ERA. That's not a pitching mismatch, that's a pitching execution. The only question is whether -186 is too much juice. Claude found a cheaper line at -150.
San Francisco is the sneaky consensus play. Grok takes Giants ML at +102, ChatGPT has them at -108 for 2.5u (his top play), and both Claude and Gemini hit the SF/CIN Over. Robbie Ray against Brady Singer's 7.71 ERA at Great American Ball Park is the matchup driving all of it. Four AIs, four plays involving the same game, four different approaches.
The Orioles create an AI head-to-head clash. Grok (BAL -149, 1.5u), ChatGPT (BAL -145, 1.5u), and Claude (BAL -150, 2u) all back Baltimore. But Gemini goes the other way with Arizona ML at +122 for 2u, arguing Merrill Kelly's FIP is actually better than Trevor Rogers' underlying numbers. Three against one. Gemini is playing contrarian again.
Claude loads the heaviest card at 18.5u across 10 plays. Three totals unders (TOR/MIL, TEX/ATH, SEA/SD), three F5 plays, and four side bets. It's the most aggressive Claude board of the season. ChatGPT commits 9u on 5 plays, and his Angels ML at +153 is the day's biggest dog shot, arguing Reid Detmers' xERA tells a different story than his surface numbers.
Texas-Oakland draws multi-AI action from completely different angles. Gemini wants Rangers ML at -140 for 2.5u plus the F5 ML. Claude wants the Under. MacKenzie Gore and Jeffrey Springs on the mound means runs could be scarce, or Texas could simply roll. Different thesis, same game.
Claude is the only AI in the contest that consistently plays run lines, and Monday showed exactly why. Four run line sweeps in a single day (Seattle, Baltimore, Philadelphia, Texas) turned what would have been a decent 5-4 ML day into a dominant 9-4 performance. The plus-money on those -1.5 lines adds up fast when they hit.
But here's the other side: Claude's overall record at 53-63-7 is the second worst in the contest. The run line aggression means more losses on days when favorites win tight. A team winning 3-2 gives you the ML but kills the -1.5. Claude's strategy is high variance by design, and it requires blowout days like Monday to make the math work.
Through 123 graded picks, Claude leads the contest in total volume by a wide margin. More picks means more exposure, more juice paid on losses, and a deeper hole when cold streaks hit. But days like April 13, where the sweeps cascade into +5.10u in a single night, are exactly what the model is built for. Feast or famine. Monday was a feast.