DuelLabGameBench 2
Menu

Model results · OpenAI

GPT-5.4 Mini

No Overall rank: results for 2 of 3 reasoning settings. Missing Intensive.

Score · relative 0–100
38.9
Rank
—
Status
Provisional
Players generated
2026-07-02
Match evidence updated
2026-10-26
Results published
Compare in charts View player programs Back to leaderboard

Scores by reasoning setting

  • Medium (Balanced group, tested and included)48.3
  • None (Baseline group, tested and included)29.5

Every tested setting is included in the overall score.

Available settings and grouping

Available settings

NoneLowMediumHighXHigh

Checked on 2026-07-21 · Provider documentation

How settings are grouped

  • Baseline: None — tested and included
  • Balanced: Medium — tested and included
  • Intensive: XHigh — available, not tested in the current results

Scores by game

  • Go51.8
  • Chess41.5
  • Shogi45.5
  • Xiangqi43.9
  • Hex29.5
  • International Draughts39.3
  • Game of the Amazons26.2
  • Tumbleweed33.5

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: Medium (48.3).

Across included settings, the highest game score is Go (51.8); the lowest is Game of the Amazons (26.2). 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
gpt-5.4-mini
Provider route (catalog)
openai
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 24 indexed attempts)

Generation evidence

  • None: 8 contributing player programs across 8 games
  • Medium: 8 contributing player programs across 8 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)
GoBalanced1 working of 1 attempt69.71668.7331 / 36 / 149
GoBaseline1 working of 1 attempt33.91312.686 / 159 / 271
ChessBalanced1 working of 1 attempt41.51406.39 / 365 / 106
ChessBaseline1 working of 1 attempt41.51406.31 / 381 / 98
ShogiBalanced1 working of 1 attempt45.81390.8124 / 166 / 204
ShogiBaseline1 working of 1 attempt45.31386.314 / 381 / 99
XiangqiBalanced1 working of 1 attempt44.31430.551 / 322 / 125
XiangqiBaseline1 working of 1 attempt43.51425.064 / 289 / 145
HexBalanced1 working of 1 attempt48.01412.5246 / 0 / 284
HexBaseline1 working of 1 attempt11.1821.154 / 0 / 476
International DraughtsBalanced1 working of 1 attempt47.71384.1205 / 24 / 273
International DraughtsBaseline1 working of 1 attempt31.01131.7130 / 10 / 362
Game of the AmazonsBalanced1 working of 1 attempt45.91397.0230 / 0 / 300
Game of the AmazonsBaseline1 working of 1 attempt6.4719.528 / 0 / 502
TumbleweedBalanced1 working of 1 attempt43.31313.5152 / 128 / 258
TumbleweedBaseline1 working of 1 attempt23.71098.823 / 188 / 327

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

0 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

Medium: 88% / 12% / 0% · n 8None: 75% / 25% / 0% · n 8 No Overall rankMissing Intensive

Generation cost

$0.22average estimated cost per game and reasoning setting

Median
$0.17
Range
$0.01–$1.50
Known cost subtotal (incomplete or unverified charges)
$5.17
Cost data
17 combinations · 18 runs
Price source
Combined: Recorded attempt + Standard list-price estimate
Price list
boardgame-list-prices-2026-07-12 + boardgame-list-prices-2026-10-09
Average output
61k tokens
Output range
1.6k–384k tokens
Runs with output-token data
18
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