DuelLabGameBench 2
Menu

Model results · OpenAI

GPT-5.4 Nano

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

Score · relative 0–100
29.5
Rank
—
Status
Provisional
Players generated
2026-07-02 – 2026-07-09
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)32.5
  • None (Baseline group, tested and included)26.6

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

  • Go31.3
  • Chess1 of 4 failed30.2
  • Shogi3 of 4 failed8.7
  • Xiangqi3 of 4 failed12.9
  • Hex29.5
  • International Draughts50.9
  • Game of the Amazons1 of 4 failed22.9
  • Tumbleweed49.8

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 (32.5).

Across included settings, the highest game score is International Draughts (50.9); the lowest is Shogi (8.7). 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-nano
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 57 indexed attempts)

Generation evidence

  • None: 12 contributing player programs across 7 games
  • Medium: 12 contributing player programs across 7 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)
GoBalanced2 working of 2 attempts16.2 program scores 25.8 · 6.71059.9 program range 895–122575 / 161 / 796
GoBaseline2 working of 2 attempts46.4 program scores 52.1 · 40.61434.4 program range 1379–1489346 / 223 / 463
ChessBalanced1 working of 2 attempts40.31396.32 / 367 / 111
ChessBaseline2 working of 2 attempts40.3 program scores 40.5 · 40.21396.8 program range 1396–139812 / 719 / 229
ShogiBaseline1 working of 2 attempts34.61293.083 / 142 / 269
XiangqiBalanced1 working of 2 attempts51.61487.1129 / 238 / 131
HexBalanced2 working of 2 attempts50.7 program scores 82.9 · 18.51488.8 program range 983–1995526 / 0 / 534
HexBaseline2 working of 2 attempts8.4 program scores 8.4 · 8.4746.1 program range 746–74680 / 0 / 980
International DraughtsBalanced2 working of 2 attempts60.2 program scores 63.7 · 56.71583.2 program range 1525–1641548 / 35 / 421
International DraughtsBaseline2 working of 2 attempts41.6 program scores 64.3 · 18.81296.9 program range 941–1652384 / 2 / 618
Game of the AmazonsBalanced2 working of 2 attempts35.1 program scores 60.3 · 9.91206.2 program range 832–1580354 / 0 / 706
Game of the AmazonsBaseline1 working of 2 attempts21.21063.0101 / 0 / 429
TumbleweedBalanced2 working of 2 attempts51.4 program scores 76.5 · 26.41472.6 program range 1130–1815481 / 93 / 502
TumbleweedBaseline2 working of 2 attempts48.2 program scores 73.7 · 22.71422.0 program range 1087–1757398 / 189 / 489

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

8 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: 31% / 44% / 25% · n 16None: 31% / 44% / 25% · n 16 No Overall rankMissing IntensiveMany programs failed

Generation cost

$0.03average estimated cost per game and reasoning setting

Median
$0.02
Range
$0.0049–$0.27
Known cost subtotal (incomplete or unverified charges)
$1.08
Cost data
17 combinations · 34 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
22k tokens
Output range
2.7k–256k tokens
Runs with output-token data
34
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