Today's Picks: April 4, 2026
18 picks across 4 AI models. Mariners ML and Blue Jays ML are 3/4 consensus plays.
Day 2 Consensus Board
Padres ML — 2/4 Models ✗ LOSS
Models: ChatGPT (1.5u), Grok (1u)
Tigers upset San Diego 5-2. Framber Valdez outdueled Michael King in his Detroit debut.
Dodgers ML — 2/4 Models ✓ WIN
Models: ChatGPT (1u), Grok (1u)
Los Angeles held on 5-4 over Arizona. Both models cash on the heavy chalk.
Blue Jays ML — 2/4 Models ✓ WIN
Models: Gemini (2u), Grok (1u)
Toronto edged Oakland 3-2. Gausman delivered as expected in the home opener.
- Braves ML ✓ — Grok only (1u) • ATL 6, KC 0
- Mariners ML ✓ — Grok only (1u) • SEA 5, CLE 1
- Marlins ML ✓ — Gemini only (3u) • MIA 2, COL 1
Day 2 Takeaways
- Gemini goes 2-for-2 and takes the lead at +11.50u total. Both high-conviction plays (3u Marlins, 2u Blue Jays) cashed.
- Grok continues its volume approach with 5 picks, going 4-1. The only miss was the Padres consensus play.
- Padres ML was the only losing consensus. Framber Valdez's Tigers debut spoiled both ChatGPT and Grok.
- Claude sat out Day 2, preserving its 6-1 record. Selective play was a hallmark of its 2025 championship run.
- All four models are profitable through 2 days. The field combined for 27-7-1 overall.
ⓘ How Units Are Graded
The "units" column = what you're trying to win. Favorite win = +units. Favorite loss = -units x (|odds|/100), because you risked more. Underdog win = +units x (odds/100), because you win more. Underdog loss = -units. Example: 1u on -110 wins +1.00u, loses -1.10u.
Daily Scorecards
Most recent day first
Day 2 — Friday, March 27
Day 1 — Opening Day, March 26
All picks through Day 2 (March 27) have been graded. Gemini leads the field at +11.50 units.
Meet the Contenders
The defending champion faces three challengers, including a brand-new competitor. Each AI model brings a different analytical approach to handicapping MLB games.
The 2025 champion dominated with the highest ROI in the field. Known for conservative, high-conviction selections and strong situational awareness. Returns to defend the crown over a full season.
Led the field in total picks during 2025 and posted a profitable season but trailed Claude in ROI. The most aggressive picker in the group with the highest total volume.
Finished in the red during the 2025 contest despite a winning record, highlighting the importance of line value. Looking to prove the doubters wrong over a full 162-game season.
The wild card entry for 2026. Built by Elon Musk's xAI, Grok brings a different training approach and real-time data access through the X platform. No track record yet. Anything can happen.
What's New in 2026
Bigger, longer, more bet types, and a new challenger. Here's how Season 2 builds on a dominant 2025 where all three AI models posted winning records.
📅 Full Season Coverage
The 2025 contest ran 24 days. The 2026 showdown covers the entire MLB regular season from Opening Day on March 26 through the end of September. No more small sample sizes. This is a true long-haul test of AI handicapping ability.
📄 Daily Picks, Daily Results
Every contestant's picks are posted daily before first pitch. Come back each day to see what Claude, ChatGPT, Gemini, and Grok are betting, then check the results page to see exactly how every pick landed. Full transparency, nothing hidden.
🤖 Four Models, Not Three
Grok by xAI joins the battle as a brand new challenger. In 2025, all three AI models finished with winning records and Claude dominated at +46.53 units profit. Can Grok hang with proven winners? Can anyone dethrone the defending champ?
🎲 Multiple Bet Types
2025 was moneylines only. 2026 expands to moneylines, run lines, and totals (over/under). Each model picks the bet type that offers the best edge on any given game. Props are excluded to keep verification clean and accurate.
Contest Rules
Moneylines
Straight win picks
Run Lines
Spread bets (-1.5 / +1.5)
Totals
Over / Under runs
Props
Not included
Variable Unit Sizing
Each model sizes bets from 0.5 to 3 units based on confidence. Units represent what you're trying to win. Favorite losses cost more (risk-based grading). This rewards conviction and punishes overconfidence.
Picks Posted Pre-Game
All picks are published before first pitch. No retroactive additions, no editing after games start. Time stamps on every selection.
Complete Transparency
Every pick from every model is tracked with full history. Wins, losses, and pushes are all public. No cherry-picking, no hidden results.
Variable Daily Picks
The number of picks per model varies based on the daily slate. More games means more potential selections. Each AI decides its own volume based on edge.
All Odds at Time of Pick
Picks are recorded with the odds available at the time of selection. No "best line" shopping or retroactive line adjustments.
Monthly & Season Standings
Live leaderboard updated daily with monthly breakdowns. Track which model runs hot in April versus who dominates the dog days of August.
2025 Season Recap: The AIs Delivered
The inaugural AI Showdown proved that artificial intelligence can genuinely pick baseball. All three models finished with winning records, and the overall results were remarkable for a first-ever AI handicapping contest.
Claude by Anthropic
All three models beat 50% win rate. Two of three were profitable. Claude crushed it. Now they run it back for 162 games with a new challenger.
Can Anyone Dethrone Claude?
All four AI models post their picks daily before first pitch. Check back every day to see the selections and track the results in real time.
Back to Home