CHESS BENCHMARK · REASONING MEDIUM

Gemini 4 Carbon

Google · released 2026-10-28 · route gemini-4-carbon · Model profile

Opening gap: +0 (95%: -1,146 to +1,146). The interval includes zero, so we can't yet say whether it performs differently across the variants.

Standard

Rating
1,41495%: 604 to 2,225
Record
1 won, 0 drawn, 0 lost of 1
Accuracy (open positions)
96.7
Accuracy (all moves)
96.3
Average centipawn loss
22
Blunders / mistakes / inaccuracies per 100 moves
0.0 / 0.0 / 0.0
Matched Stockfish's first choice
63%
In Stockfish's top three
81%
Legal on the first try
100.0%
Illegal / invalid
0 of 27 answers
Forfeits
0
Adjudicated games
0
API-equivalent cost per game
$0.493
API-equivalent cost per move
$0.018
Seconds per move
25.3
Model minutes per game
11.4
Tokens per move (in / out)
2,240 / 1,378
Reasoning tokens per move
1,365
Aborted on our side (not counted)
0

By phase

PhaseMovesAccuracyCentipawn lossBlunders /100Engine match
Opening1098.240.080%
Middlegame1695.4350.050%
Endgame1n/a00.0100%

Chess960

Rating
1,41495%: 604 to 2,225
Record
1 won, 0 drawn, 0 lost of 1
Accuracy (open positions)
91.0
Accuracy (all moves)
92.2
Average centipawn loss
32
Blunders / mistakes / inaccuracies per 100 moves
3.6 / 3.6 / 7.1
Matched Stockfish's first choice
57%
In Stockfish's top three
68%
Legal on the first try
100.0%
Illegal / invalid
0 of 28 answers
Forfeits
0
Adjudicated games
1
API-equivalent cost per game
$1.25
API-equivalent cost per move
$0.045
Seconds per move
59.2
Model minutes per game
27.6
Tokens per move (in / out)
2,300 / 3,996
Reasoning tokens per move
3,984
Aborted on our side (not counted)
0

By phase

PhaseMovesAccuracyCentipawn lossBlunders /100Engine match
Opening1091.62910.060%
Middlegame1890.5330.056%
Endgame0n/an/an/an/a

Games