NONOGRAMS BENCHMARK · REASONING MEDIUM

Gemini 4 Argon

Google · released 2026-09-30 · route gemini-4-argon · Model profile · Chess results · Mines results · FreeCell results · Starjar results · Nations Royale results · Hustle Sim results · Elemental Armory results · Micro Mouse results · Ravelin results

Results

Score
5095%: 28 to 72 · 6 of 36 base puzzles so far
Puzzles solved
3 of 6
Cells that match the solution
85.6%
Rows and columns whose clue it satisfies
70.0%
Wrong cells / left cells unknown / no grid
3 / 0 / 0
Near misses (three cells or fewer wrong)
0
Average stated confidence
77%
Average confidence when it had not solved it
53%
Sure but wrong (90% or more, unsolved)
1
Calibration (Brier, lower is better)
0.200
API-equivalent cost per solved puzzle
$0.469
Tokens per puzzle (in / out)
3,095 / 22,834
Reasoning tokens per puzzle
247
Retries for an unreadable reply
0
API-equivalent cost per puzzle
$0.234
Seconds per puzzle
212.4
Replies cut off at the length limit
0.0%
Aborted on our side (not counted)
6

By tier

TierNeedsSolvedOutput tokensSeconds
5 by 5Line logic, a round or two1 of 141910.2
10 by 10Line logic, about 3 rounds0 of 163513.9
15 by 15Line logic, about 5 rounds1 of 124,811212.2
20 by 20Line logic, about 8 rounds1 of 144,113397.2
25 by 25Line logic, about 11 rounds0 of 166,899631.3
BeyondLine logic stalls; needs contradiction or search0 of 112810.0

Cells by the depth of logic they need

Each cell is tagged with the round of line logic that settles it. Beyond line logic means only contradiction or search does.

DepthCellsRight
First pass40394.0%
Rounds 2-361390.0%
Rounds 4-733281.3%
Round 8 on15968.5%
Beyond line logic9363.4%

Every puzzle