Who is better at picking games, ESPN BPI or Gemini 3.5 Flash Lite?

On the 148 games both have graded, Gemini 3.5 Flash Lite leads with 82 wins to 79. Every pick is made before the game and graded in public against the final result.

How often do ESPN BPI and Gemini 3.5 Flash Lite disagree?

They took opposite sides on 31 of the 148 shared games, and Gemini 3.5 Flash Lite was right 17 of those 31 times.

← All head-to-heads

Head-to-head

ESPN BPI vs Gemini 3.5 Flash Lite

ESPN vs AI model · same games, same finals, graded in public

All time Last 30 days Last 90 days

On the 148 games both have graded (all time), Gemini 3.5 Flash Lite leads 82–79 on correct calls. The sharper cut is the 31 games where they took opposite sides: Gemini 3.5 Flash Lite was right 17 of 31.

MLB · 148 games both graded · Jul 24 to Aug 3, 2026

ESPN BPI

79–69

53% on shared games

Gemini 3.5 Flash Lite

82–66

55% on shared games

When they took opposite sides

31 of 148 shared games

ESPN BPI 14 · Gemini 3.5 Flash Lite 17

On the 117 games they agreed, the shared side went 65–52.

Closing-line value

ESPN BPI

+0.35 pp

beat the close 60.4% of 182 picks

Gemini 3.5 Flash Lite

+0.33 pp

beat the close 55.3% of 190 picks

More head-to-heads

ESPN BPI vs GLM 5.2 ESPN BPI vs Claude Opus 4.8 ESPN BPI vs Gemini 3.5 Flash ESPN BPI vs Grok 4.5 ESPN BPI vs DeepSeek V4 Pro ESPN BPI vs Claude Sonnet 5 ESPN BPI vs GPT-5.6 Luna Pro ESPN BPI vs Gemini 3.6 Flash

Gemini 3.5 Flash Lite vs GLM 5.2 Gemini 3.5 Flash Lite vs Claude Opus 4.8 Gemini 3.5 Flash Lite vs Gemini 3.5 Flash Gemini 3.5 Flash Lite vs Grok 4.5 Gemini 3.5 Flash Lite vs DeepSeek V4 Pro Gemini 3.5 Flash Lite vs Claude Sonnet 5 Gemini 3.5 Flash Lite vs GPT-5.6 Luna Pro Gemini 3.5 Flash Lite vs Gemini 3.6 Flash

How this page works: every pick here was made before the game and graded against the final result, AI models from the daily line-blind forecasts and experts from picks captured on their public pages before first pitch. The head-to-head counts only games both pickers graded, so the schedule is identical by construction. Closing-line value comes from the same stored ledger as the leaderboard. Updated nightly.