DuelLabGameBench 2
Menu

Model results · Inception

Mercury 2.5 Preview

Rank 63 of 66 in Overall Models. All three benchmark settings tested.

Score · relative 0–100
21.9
Rank
63
Status
Complete 3/3
Rank change
0
Players generated
2026-09-03
Match evidence updated
2026-10-26
Results published
Compare in charts View player programs Back to leaderboard

Scores by reasoning setting

  • High (Intensive group, tested and included)21.1
  • Medium (Balanced group, tested and included)20.3
  • None (Baseline group, tested and included)24.2

Every tested setting is included in the overall score.

Available settings and grouping

Available settings

NoneLowMediumHigh

Checked on 2026-09-03 · Provider documentation

How settings are grouped

  • Baseline: None — tested and included
  • Balanced: Medium — tested and included
  • Intensive: High — tested and included

Scores by game

  • Go2 of 3 failed9.1
  • Chess42.7
  • Shogi2 of 3 failed15.3
  • Xiangqi41.3
  • Hex1 of 3 failed18.8
  • International Draughts18.3
  • Game of the Amazons1 of 3 failed8.1
  • Tumbleweed1 of 3 failed21.5

Each game counts every generation attempt, a failed one as 0 (“1 of 2 failed” marks games with failed attempts). Scores are relative within each game, across the included settings; they do not compare game difficulty.

What this result shows

Strongest included setting: None (24.2).

Across included settings, the highest game score is Chess (42.7); the lowest is Game of the Amazons (8.1). Each game counts every generation attempt, a failed one as 0. These are relative scores within each game, not direct comparisons of difficulty.

Match evidence measures these retained programs. It does not establish how a fresh generation will perform. Independent-generation variation is not estimated in this release.

Test protocol and evidence limits
Model identifier
inception/mercury-2.5-preview
Provider route (catalog)
openrouter
Prompt identity / profile
Not retained in this result set / Not retained in this result set
Prompt version
Not retained in this result set
Generation policy
Generation-quality admission policy (see methodology)
Requested output-token limits
Not retained (0 of 37 indexed attempts)

Generation evidence

  • None: 5 contributing player programs across 5 games
  • Medium: 6 contributing player programs across 6 games
  • High: 6 contributing player programs across 6 games

Counts describe contributing programs, not independent replications: repairs and resends are not new independent samples. Independent-generation lineage is not retained.

GameReasoning groupProgramsWorking-program strengthRaw ratingCombined record (W / D / L)
GoIntensive1 working of 1 attempt27.41243.174 / 118 / 324
ChessIntensive1 working of 1 attempt42.51413.72 / 388 / 90
ChessBalanced1 working of 1 attempt43.21419.41 / 397 / 82
ChessBaseline1 working of 1 attempt42.31412.10 / 390 / 90
ShogiBaseline1 working of 1 attempt46.01392.6150 / 116 / 228
XiangqiIntensive1 working of 1 attempt39.01389.812 / 349 / 137
XiangqiBalanced1 working of 1 attempt40.81403.527 / 336 / 135
XiangqiBaseline1 working of 1 attempt44.01429.025 / 372 / 101
HexBalanced1 working of 1 attempt16.8949.384 / 0 / 446
HexBaseline1 working of 1 attempt39.71307.9203 / 0 / 327
International DraughtsIntensive1 working of 1 attempt15.1874.962 / 3 / 437
International DraughtsBalanced1 working of 1 attempt18.0927.975 / 3 / 424
International DraughtsBaseline1 working of 1 attempt21.7987.891 / 3 / 408
Game of the AmazonsIntensive1 working of 1 attempt14.0929.865 / 0 / 465
Game of the AmazonsBalanced1 working of 1 attempt10.3843.047 / 0 / 483
TumbleweedIntensive1 working of 1 attempt31.01181.857 / 194 / 287
TumbleweedBalanced1 working of 1 attempt33.41207.522 / 288 / 228

Each row covers all the model's programs in one game and setting. Strength and raw rating are attempt-weighted means; under them, each program's score and the lowest and highest rating describe the programs (not a confidence interval). The combined record adds up their matches: evidence volume, not strength.

Failures and budgets

7 model-code generation failures; 0 provider/infrastructure failures excluded from the generation denominator. Runtime player faults are separate: 0 recorded one-sided faults across contributing program rows. A crash or illegal move pinned on one program also counts as that program's loss in its match record; other faults, such as timeouts, are left out of rating evidence.

Repair/resend outcomes and available costs are included. Exact retry ceilings and execution budgets are not retained in this result set; see the methodology for the public policy. Do not infer a measured setting from a current provider catalog.

Player program results

High: 63% / 12% / 25% · n 8Medium: 75% / 0% / 25% · n 8None: 25% / 38% / 37% · n 8 All three benchmark settings testedMany programs failed

Generation cost

$0.0012average estimated cost per game and reasoning setting

Median
$0.000945
Range
$0.000440–$0.0026
Provider-reported total
$0.03
Cost data
24 combinations · 24 runs
Price source
Provider-reported
Price list
Not recorded for this older data
Average output
4.9k tokens
Output range
1.0k–11k tokens
Runs with output-token data
24
How cost is measured

Comparison cost gives equal weight to each model, game, and reasoning-setting combination. Recorded generation attempts include repairs and resends when cost evidence is available; runtime compute and service operation are excluded. Cost and output-token counts use separate telemetry denominators. A list-price estimate is not actual expenditure; mixed evidence is not a uniform standard-price comparison.

Test protocol

  • Combines this model's tested reasoning settings
  • Player-program generation; retained protocol details are under “Test protocol and evidence limits” above