Scores by reasoning setting

Enabled (Balanced group, tested and included)31.3
None (Baseline group, tested and included)31.9

Every tested setting is included in the overall score.

Available settings and grouping

Available settings

NoneEnabled

Checked on 2026-07-24 · Provider documentation

How settings are grouped

Baseline: None — tested and included
Balanced: Enabled — tested and included
Intensive: Not available for this model

Scores by game

Go1 of 4 failed35.4
Chess1 of 4 failed31.8
Shogi39.1
Xiangqi1 of 2 failed20.2
Hex38.2
International Draughts20.2
Game of the Amazons45.2
Tumbleweed29.9

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: None (31.9).

Across included settings, the highest game score is Game of the Amazons (45.2); the lowest is Xiangqi (20.2). 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: poolside/laguna-s-2.1
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: Not retained (0 of 72 indexed attempts)

Generation evidence

None: 12 contributing player programs across 7 games
Enabled: 11 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.

Game	Reasoning group	Programs	Working-program strength	Raw rating	Combined record (W / D / L)
Go	Balanced	1 working of 2 attempts	40.9	1382.3	158 / 86 / 272
Go	Baseline	2 working of 2 attempts	50.3 program scores 59.7 · 40.9	1472.7 program range 1382–1563	415 / 166 / 451
Chess	Balanced	2 working of 2 attempts	43.0 program scores 43.2 · 42.8	1417.8 program range 1416–1419	2 / 790 / 168
Chess	Baseline	1 working of 2 attempts	41.0	1402.2	18 / 342 / 120
Shogi	Balanced	2 working of 2 attempts	38.8 program scores 41.9 · 35.6	1329.5 program range 1302–1357	196 / 302 / 490
Shogi	Baseline	2 working of 2 attempts	39.4 program scores 43.1 · 35.8	1335.4 program range 1304–1367	179 / 349 / 460
Xiangqi	Balanced	1 working of 1 attempt	40.4	1400.3	44 / 298 / 156
Hex	Balanced	2 working of 2 attempts	28.1 program scores 40.1 · 16.0	1123.4 program range 934–1313	285 / 0 / 775
Hex	Baseline	2 working of 2 attempts	48.4 program scores 78.5 · 18.4	1428.8 program range 979–1878	500 / 0 / 560
International Draughts	Balanced	1 working of 1 attempt	19.4	951.0	81 / 3 / 418
International Draughts	Baseline	1 working of 1 attempt	21.1	978.8	87 / 6 / 409
Game of the Amazons	Baseline	2 working of 2 attempts	45.2 program scores 56.7 · 33.7	1387.7 program range 1242–1533	454 / 0 / 606
Tumbleweed	Balanced	2 working of 2 attempts	29.3 program scores 30.0 · 28.6	1162.3 program range 1155–1170	143 / 294 / 639
Tumbleweed	Baseline	2 working of 2 attempts	30.5 program scores 30.7 · 30.2	1175.9 program range 1173–1179	128 / 349 / 599

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

3 model-code generation failures; 5 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

Enabled: 59% / 33% / 8% · n 12None: 50% / 36% / 14% · n 14 No Overall rankIntensive not supported

Generation cost

$0.0032average estimated cost per game and reasoning setting

Median: $0.0031
Range: $0.0012–$0.0068
Known cost subtotal (incomplete or unverified charges): $0.09
Cost data: 16 combinations · 28 runs
Price source: Provider-reported
Price list: Not recorded for this older data
Average output: 8.3k tokens
Output range: 1.7k–42k tokens
Runs with output-token data: 28

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

Laguna S 2.1