NONOGRAMS BENCHMARK · REASONING MEDIUM

GPT-6 Sol

Openai · released 2026-09-22 · route gpt-6-sol · Model profile · Chess results · Sudoku results · Mines results · FreeCell results · Starjar results · Nations Royale results · Hustle Sim results · Elemental Armory results · Micro Mouse results

Results

Score
3395%: 12 to 55 · 6 of 36 base puzzles so far
Puzzles solved
2 of 6
Cells that match the solution
26.9%
Rows and columns whose clue it satisfies
18.9%
Wrong cells / left cells unknown / no grid
0 / 4 / 0
Near misses (three cells or fewer wrong)
0
Average stated confidence
38%
Average confidence when it had not solved it
6%
Sure but wrong (90% or more, unsolved)
0
Calibration (Brier, lower is better)
0.010
API-equivalent cost per solved puzzle
$0.213
Tokens per puzzle (in / out)
29,739 / 5,410
Reasoning tokens per puzzle
4,681
Retries for an unreadable reply
0
API-equivalent cost per puzzle
$0.071
Seconds per puzzle
100.1
Replies cut off at the length limit
0.0%
Aborted on our side (not counted)
0

By tier

TierNeedsSolvedOutput tokensSeconds
5 by 5Line logic, a round or two1 of 12076.8
10 by 10Line logic, about 3 rounds1 of 15,09999.0
15 by 15Line logic, about 5 rounds0 of 19,381179.1
20 by 20Line logic, about 8 rounds0 of 16,368117.7
25 by 25Line logic, about 11 rounds0 of 17,055117.2
BeyondLine logic stalls; needs contradiction or search0 of 14,35080.6

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 pass40378.4%
Rounds 2-361316.8%
Rounds 4-73323.6%
Round 8 on1590.0%
Beyond line logic930.0%

Every puzzle