Who is better at picking games, Grok 4.5 or Kimi K3?

On the 308 games both have graded, Grok 4.5 leads with 178 wins to 175. Every pick is made before the game and graded in public against the final result.

How often do Grok 4.5 and Kimi K3 disagree?

They took opposite sides on 21 of the 308 shared games, and Grok 4.5 was right 12 of those 21 times.

← All head-to-heads

Head-to-head

Grok 4.5 vs Kimi K3

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

All time Last 30 days Last 90 days

On the 308 games both have graded (all time), Grok 4.5 leads 178–175 on correct calls. The sharper cut is the 21 games where they took opposite sides: Grok 4.5 was right 12 of 21.

MLB · 270 games both graded · Jul 17 to Aug 6, 2026

Grok 4.5

151–119

56% on shared games

Kimi K3

147–123

54% on shared games

When they took opposite sides

20 of 270 shared games

Grok 4.5 12 · Kimi K3 8

On the 250 games they agreed, the shared side went 139–111.

UFC · 38 games both graded · Jul 18 to Aug 1, 2026

Grok 4.5

27–11

71% on shared games

Kimi K3

28–10

74% on shared games

When they took opposite sides

1 of 38 shared games

Grok 4.5 0 · Kimi K3 1

On the 37 games they agreed, the shared side went 27–10.

Closing-line value

Grok 4.5

+0.34 pp

beat the close 56.8% of 331 picks

Kimi K3

+0.36 pp

beat the close 55.9% of 270 picks

More head-to-heads

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

Kimi K3 vs GLM 5.2 Kimi K3 vs Claude Opus 4.8 Kimi K3 vs Gemini 3.5 Flash Kimi K3 vs DeepSeek V4 Pro Kimi K3 vs Claude Sonnet 5 Kimi K3 vs GPT-5.6 Luna Pro Kimi K3 vs Gemini 3.5 Flash Lite Kimi K3 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.