Score · relative 0–100
68.2
Rank
20
Status
Partial 3/3
Rank change
New
Players generated
2026-11-16
Match evidence updated
2026-11-19

Why this result is partial: some games need a new attempt: their only generation attempts failed because of our test setup or the provider, not the model. The rank uses the available reasoning-setting results.

Reasoning settings compared

  1. XHighIntensive

    75.2 · no rank on its setting leaderboard

    Cost per program
    $0.189 · provider-reported
    Working programs
    7 of 7
    API time
    13.3 min (observed mean, descriptive)

    7.3 points below High, at about the same cost

    Not compared: no official rank on its setting leaderboard; its score does not cover every game in the cost comparison.

    • No alternatives: this setting is not compared (no official rank on its setting leaderboard).
  2. HighBalanced

    82.5 · rank 12 on the Balanced leaderboard

    Cost per program
    $0.181 · provider-reported
    Working programs
    8 of 8
    API time
    11.6 min (observed mean, descriptive)

    35.8 points above None, at 2.9× the cost

    • Higher score for $0.181 or less: GPT-6.2 Sol Medium · Balanced: 93.4 for $0.0972 (list-price estimate)
    • Cheapest scoring at least as high: Pareto 26.10 (no Overall rank) Reasoning controls unavailable · Baseline: 89.9 for $0.0296 (provider-reported) (one of 9 such settings)
  3. NoneBaseline

    46.7 · rank 43 on the Baseline leaderboard

    Cost per program
    $0.0631 · provider-reported
    Working programs
    6 of 8
    API time
    2.3 min (observed mean, descriptive)
    • Higher score for $0.0631 or less: Pareto 26.10 (no Overall rank) Reasoning controls unavailable · Baseline: 89.9 for $0.0296 (provider-reported)
    • Cheapest scoring at least as high: Muse Ember Minimal · Baseline: 46.9 for $0.00281 (provider-reported) (one of 42 such settings)

Score and cost of every ranked setting

The chart needs JavaScript. The setting cards hold this model's results, and Charts compares every model.

Each setting is placed by its published score and its generation cost per player program, the mean over the game cells including repairs and failed attempts. Higher is better; further left costs less (logarithmic axis, with recorded $0 in its own lane). Filled marks are compared; hollow marks are not (the reason is in the table). The line marks the best value across ranked reasoning settings: no ranked setting with complete cost evidence scores at least as high for the same or less. It includes models without an Overall rank, so it can differ from the line on Charts. Costs mix provider-reported charges and list-price estimates; differences in score points are not tested for significance.
Show the numbers

The table needs JavaScript.

Available settings and grouping

Available settings

NoneLowHighXHigh

Checked on 2026-11-16 · Provider documentation

How settings are grouped

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

Scores by game

  • Go71.3
  • Chess1 of 3 failed60.6
  • Shogi85.6
  • Xiangqi60.8
  • Hex87.4
  • International Draughts1 of 3 failed44.8
  • Game of the Amazons72.3
  • Tumbleweed61.1

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: High (82.5).

Across included settings, the highest game score is Hex (87.4); the lowest is International Draughts (44.8). 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
qwen/qwen4-max
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
131072 (28 of 28 indexed attempts)

Generation evidence

  • None: 6 contributing player programs across 6 games
  • High: 8 contributing player programs across 8 games
  • XHigh: 7 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)
GoBalanced1 working of 1 attempt61.71561.8364 / 23 / 263
GoBaseline1 working of 1 attempt80.81791.1472 / 64 / 116
ChessIntensive1 working of 1 attempt92.41954.2517 / 54 / 39
ChessBalanced1 working of 1 attempt89.51881.6492 / 65 / 53
ShogiIntensive1 working of 1 attempt91.62019.5554 / 0 / 80
ShogiBalanced1 working of 1 attempt90.81989.3546 / 4 / 84
ShogiBaseline1 working of 1 attempt74.41635.5419 / 18 / 197
XiangqiIntensive1 working of 1 attempt83.91784.5452 / 100 / 76
XiangqiBalanced1 working of 1 attempt60.11526.7292 / 112 / 224
XiangqiBaseline1 working of 1 attempt38.51353.444 / 352 / 232
HexIntensive1 working of 1 attempt86.62081.5544 / 0 / 118
HexBalanced1 working of 1 attempt94.32408.6603 / 0 / 59
HexBaseline1 working of 1 attempt81.31907.7504 / 0 / 160
International DraughtsIntensive1 working of 1 attempt39.81234.5215 / 7 / 410
International DraughtsBalanced1 working of 1 attempt94.52321.7579 / 20 / 35
Game of the AmazonsIntensive1 working of 1 attempt88.62087.4560 / 0 / 102
Game of the AmazonsBalanced1 working of 1 attempt88.92102.2563 / 0 / 99
Game of the AmazonsBaseline1 working of 1 attempt39.41226.0224 / 0 / 438
TumbleweedIntensive1 working of 1 attempt43.81235.7266 / 0 / 394
TumbleweedBalanced1 working of 1 attempt80.21967.5507 / 0 / 153
TumbleweedBaseline1 working of 1 attempt59.41510.3366 / 0 / 294

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

2 model-code generation failures; 1 provider/infrastructure failures excluded from the generation denominator. Runtime player faults are separate: 0 recorded one-sided faults across contributing program rows. A crash, an illegal move, or a timeout repeated by a later quiet replay, pinned on one program, also counts as that program's loss in its match record; other faults 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

XHigh: 100% / 0% / 0% · n 7High: 100% / 0% / 0% · n 8None: 50% / 25% / 25% · n 8 All three benchmark settings testedMany programs failed

Generation cost, all settings combined

$0.14average estimated cost per game and reasoning setting

Median
$0.16
Range
$0.00–$0.28
Provider-reported total
$3.47
Cost data
24 combinations · 24 runs
Price source
Provider-reported
Price list
boardgame-list-prices-2026-11-12
Average output
30k tokens
Output range
4.2k–59k 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