DuelLabGameBench 2
Menu

Model results · Thinking Machines

Inkling

Rank 61 of 66 in Overall Models. All three benchmark settings tested.

Score · relative 0–100
26.7
Rank
61
Status
Complete 3/3
Rank change
0
Players generated
2026-07-24 – 2026-10-15
Match evidence updated
2026-10-26
Results published
Compare in charts View player programs Back to leaderboard

Scores by reasoning setting

  • Max (Intensive group, tested and included)33.1
  • Medium (Balanced group, tested and included)19.2
  • None (Baseline group, tested and included)27.7

Every tested setting is included in the overall score.

Available settings and grouping

Available settings

NoneMinimalLowMediumHighMax

Checked on 2026-07-24 · Provider documentation

How settings are grouped

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

Scores by game

  • Go26.0
  • Chess2 of 5 failed21.3
  • Shogi44.7
  • Xiangqi1 of 5 failed35.0
  • Hex19.3
  • International Draughts2 of 5 failed21.6
  • Game of the Amazons1 of 5 failed19.8
  • Tumbleweed2 of 5 failed25.7

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: Max (33.1).

Across included settings, the highest game score is Shogi (44.7); the lowest is Hex (19.3). 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
thinkingmachines/inkling
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
262144 (20 of 112 indexed attempts)

Generation evidence

  • None: 12 contributing player programs across 8 games
  • Medium: 6 contributing player programs across 6 games
  • Max: 14 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)
GoIntensive2 working of 2 attempts40.1 program scores 49.7 · 30.51371.5 program range 1277–1466253 / 282 / 497
GoBalanced1 working of 1 attempt17.91122.055 / 61 / 400
GoBaseline2 working of 2 attempts20.0 program scores 22.4 · 17.61150.5 program range 1118–118349 / 286 / 697
ChessIntensive2 working of 2 attempts41.9 program scores 42.0 · 41.91409.2 program range 1409–141010 / 753 / 197
ChessBaseline1 working of 2 attempts44.01425.25 / 396 / 79
ShogiIntensive2 working of 2 attempts47.9 program scores 48.1 · 47.71409.3 program range 1408–1411329 / 211 / 448
ShogiBalanced1 working of 1 attempt37.11314.816 / 299 / 180
ShogiBaseline2 working of 2 attempts49.3 program scores 62.1 · 36.41424.5 program range 1309–1540319 / 262 / 407
XiangqiIntensive1 working of 2 attempts44.01429.053 / 316 / 129
XiangqiBalanced1 working of 1 attempt41.71410.735 / 329 / 134
XiangqiBaseline2 working of 2 attempts41.3 program scores 43.1 · 39.41407.5 program range 1393–142280 / 630 / 286
HexIntensive2 working of 2 attempts16.2 program scores 16.8 · 15.7937.6 program range 926–949162 / 0 / 898
HexBalanced1 working of 1 attempt14.1893.370 / 0 / 460
HexBaseline2 working of 2 attempts27.5 program scores 32.4 · 22.61131.9 program range 1055–1208279 / 0 / 781
International DraughtsIntensive1 working of 2 attempts53.81478.3244 / 9 / 249
International DraughtsBalanced1 working of 1 attempt24.11026.7102 / 3 / 397
International DraughtsBaseline1 working of 2 attempts27.41077.3111 / 14 / 376
Game of the AmazonsIntensive2 working of 2 attempts28.8 program scores 29.1 · 28.51176.0 program range 1172–1180279 / 0 / 781
Game of the AmazonsBalanced1 working of 1 attempt18.81022.789 / 0 / 441
Game of the AmazonsBaseline1 working of 2 attempts23.61100.5113 / 0 / 417
TumbleweedIntensive2 working of 2 attempts40.8 program scores 58.1 · 23.41295.1 program range 1095–1495247 / 324 / 505
TumbleweedBaseline1 working of 2 attempts72.71738.4365 / 20 / 153

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; 6 provider/infrastructure failures excluded from the generation denominator. Runtime player faults are separate: 1 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

Max: 56% / 31% / 13% · n 16Medium: 50% / 25% / 25% · n 8None: 44% / 31% / 25% · n 16 All three benchmark settings testedMany programs failed

Generation cost

$0.10average estimated cost per game and reasoning setting

Median
$0.04
Range
$0.02–$0.26
Known cost subtotal (incomplete or unverified charges)
$5.74
Cost data
24 combinations · 46 runs
Price source
Provider-reported
Price list
boardgame-list-prices-2026-10-09
Average output
27k tokens
Output range
1.4k–80k tokens
Runs with output-token data
46
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