CHESS BENCHMARK · REASONING MEDIUM

Gemini 4 Neon

Google · released 2026-11-18 · route gemini-4-neon · Model profile

Opening gap: +489 (95%: -254 to +1,232). The interval includes zero, so we can't yet say whether it performs differently across the variants.

Standard

Rating
1,81295%: 1,228 to 2,397
Record
3 won, 0 drawn, 0 lost of 3
Accuracy (open positions)
97.5
Accuracy (all moves)
98.3
Average centipawn loss
9
Blunders / mistakes / inaccuracies per 100 moves
0.0 / 0.0 / 0.0
Matched Stockfish's first choice
58%
In Stockfish's top three
92%
Legal on the first try
100.0%
Illegal / invalid
0 of 66 answers
Forfeits
0
Adjudicated games
2
API-equivalent cost per game
$0.892
API-equivalent cost per move
$0.041
Seconds per move
24.4
Model minutes per game
9.0
Tokens per move (in / out)
2,227 / 2,257
Reasoning tokens per move
2,248
Aborted on our side (not counted)
2

By phase

PhaseMovesAccuracyCentipawn lossBlunders /100Engine match
Opening3097.280.070%
Middlegame3698.7100.047%
Endgame0n/an/an/an/a

Chess960

Rating
1,32495%: 865 to 1,782
Record
2 won, 0 drawn, 1 lost of 3
Accuracy (open positions)
82.5
Accuracy (all moves)
86.2
Average centipawn loss
68
Blunders / mistakes / inaccuracies per 100 moves
2.5 / 9.9 / 13.6
Matched Stockfish's first choice
38%
In Stockfish's top three
62%
Legal on the first try
98.8%
Illegal / invalid
1 of 82 answers
Forfeits
0
Adjudicated games
1
API-equivalent cost per game
$2.72
API-equivalent cost per move
$0.101
Seconds per move
60.2
Model minutes per game
27.1
Tokens per move (in / out)
2,348 / 6,242
Reasoning tokens per move
6,232
Aborted on our side (not counted)
3

By phase

PhaseMovesAccuracyCentipawn lossBlunders /100Engine match
Opening3084.6593.343%
Middlegame5180.2732.035%
Endgame0n/an/an/an/a

Games