DuelLabGameBench 2
Menu

Model results · Other

Beam 1.0 501B A23B

Rank 44 of 66 in Overall Models. All three benchmark settings tested.

Score · relative 0–100
44.7
Rank
44
Status
Complete 3/3
Rank change
0
Players generated
2026-10-20
Match evidence updated
2026-10-24
Results published
Compare in charts View player programs Back to leaderboard

Scores by reasoning setting

  • XHigh (Intensive group, tested and included)55.6
  • Medium (Balanced group, tested and included)37.7
  • Low (Baseline group, tested and included)40.9

Every tested setting is included in the overall score.

Available settings and grouping

Available settings

LowMediumHighXHighMax

Checked on 2026-10-20 · Provider documentation

How settings are grouped

  • Baseline: Low — tested and included
  • Balanced: Medium — tested and included
  • Intensive: XHigh — tested and included

Scores by game

  • Go1 of 3 failed22.6
  • Chess43.2
  • Shogi53.9
  • Xiangqi47.0
  • Hex52.1
  • International Draughts1 of 3 failed43.5
  • Game of the Amazons26.9
  • Tumbleweed68.6

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: XHigh (55.6).

Across included settings, the highest game score is Tumbleweed (68.6); the lowest is Go (22.6). 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
reflection/beam-1.0-501b-a23b
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 (36 of 36 indexed attempts)

Generation evidence

  • Low: 7 contributing player programs across 7 games
  • Medium: 7 contributing player programs across 7 games
  • XHigh: 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)
GoIntensive1 working of 1 attempt24.91214.253 / 135 / 328
GoBaseline1 working of 1 attempt43.01402.3107 / 209 / 200
ChessIntensive1 working of 1 attempt50.21473.170 / 325 / 85
ChessBalanced1 working of 1 attempt33.91345.115 / 281 / 184
ChessBaseline1 working of 1 attempt45.51437.432 / 357 / 91
ShogiIntensive1 working of 1 attempt78.71745.3356 / 38 / 100
ShogiBalanced1 working of 1 attempt39.31334.587 / 178 / 229
ShogiBaseline1 working of 1 attempt43.81373.7143 / 109 / 242
XiangqiIntensive1 working of 1 attempt61.11563.6196 / 199 / 103
XiangqiBalanced1 working of 1 attempt40.11397.937 / 309 / 152
XiangqiBaseline1 working of 1 attempt39.81395.531 / 318 / 149
HexIntensive1 working of 1 attempt53.21476.9273 / 0 / 257
HexBalanced1 working of 1 attempt41.11325.4210 / 0 / 320
HexBaseline1 working of 1 attempt61.91591.1319 / 0 / 211
International DraughtsIntensive1 working of 1 attempt74.71841.3348 / 28 / 126
International DraughtsBalanced1 working of 1 attempt55.71508.7253 / 11 / 237
Game of the AmazonsIntensive1 working of 1 attempt39.21312.9194 / 0 / 336
Game of the AmazonsBalanced1 working of 1 attempt31.41211.7153 / 0 / 377
Game of the AmazonsBaseline1 working of 1 attempt10.1837.546 / 0 / 484
TumbleweedIntensive1 working of 1 attempt62.81564.2319 / 0 / 219
TumbleweedBalanced1 working of 1 attempt60.31525.4305 / 0 / 233
TumbleweedBaseline1 working of 1 attempt82.71955.5434 / 0 / 104

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; 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

XHigh: 63% / 37% / 0% · n 8Medium: 63% / 25% / 12% · n 8Low: 38% / 50% / 12% · n 8 All three benchmark settings testedMany programs needed repair

Generation cost

$0.02average estimated cost per game and reasoning setting

Median
$0.01
Range
$0.0041–$0.05
Provider-reported total
$0.38
Cost data
24 combinations · 24 runs
Price source
Provider-reported
Price list
boardgame-list-prices-2026-10-09
Average output
13k tokens
Output range
2.2k–44k 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