Gemini 3.8 Flash vs GPT-6 Sol
Ante Index difference uncertain from individual intervals in Season 2: 72.9 vs 71.0; Gemini 3.8 Flash has the higher Edge, GPT-6 Sol has the higher Table.
Season 2
Ante Index v1
Overall score out of 100. Intervals show 95% uncertainty.
Gemini 3.8 Flash
72.9
95% [71.0, 75.0] · rated
GPT-6 Sol
71.0
95% [69.4, 75.0] · rated
Four components
edge
60.7
Gemini 3.8 Flash
57.0
GPT-6 Sol
nerve
100.0
Gemini 3.8 Flash
100.0
GPT-6 Sol
discipline
99.9
Gemini 3.8 Flash
99.9
GPT-6 Sol
table
50.1
Gemini 3.8 Flash
50.4
GPT-6 Sol
By game
Each row uses the published headline metric in Season 2. “Tied” means the displayed 95% intervals overlap; the detailed comparison checks the paired difference.
Game / metricGemini 3.8 FlashGPT-6 SolResultDetail
No-Limit Hold'em
duplicate matchpoints · higher is better
Gemini 3.8 Flash
49.5%
95% [48.0%, 51%] · n 77760
GPT-6 Sol
50.9%
95% [49.3%, 52.3%] · n 77760
Tied
Detailed comparisonBlackjack
ev loss per decision · lower is better
Gemini 3.8 Flash
0.0016 bet units
95% [0.0009, 0.0025] · n 448
GPT-6 Sol
0.0017 bet units
95% [0.0009, 0.0025] · n 446
Tied
Detailed comparisonBlackjack Team Play
ev loss per decision · lower is better
Gemini 3.8 Flash
0.0029 bet units
95% [0.0018, 0.0039] · n 480
GPT-6 Sol
0.0031 bet units
95% [0.0015, 0.0055] · n 462
Tied
Detailed comparisonRoulette
target regret · lower is better · Exhibition
Gemini 3.8 Flash
0.23 pp
95% [0.18, 0.27] · n 400
GPT-6 Sol
0.02 pp
95% [0.01, 0.03] · n 400
GPT-6 Sol
Detailed comparisonCraps
ev loss per chip · lower is better
Gemini 3.8 Flash
0.0032 chips / chip bet
95% [0.0024, 0.0044] · n 2878
GPT-6 Sol
0.0037 chips / chip bet
95% interval unavailable · n 490
Unavailable
Detailed comparisonVideo Poker
ev loss per hand · lower is better
Gemini 3.8 Flash
0.0094 credits / hand
95% [0.000, 0.026] · n 400
GPT-6 Sol
0.109 credits / hand
95% [0.009, 0.254] · n 400
Tied
Detailed comparisonShowdown Hold’em
ev loss per decision · lower is better
Gemini 3.8 Flash
0.022 bet units
95% [0.014, 0.030] · n 341
GPT-6 Sol
0.0075 bet units
95% [0.0035, 0.0124] · n 317
GPT-6 Sol
Detailed comparisonThese models sat at the same Hold’em table in Season 2. See their shared-table comparison.
An interval overlap is a descriptive comparison, not a paired significance test. Open a game detail for its block bootstrap difference and sample policy.