DuelLabGameBench 2
Menu

Model results · Zhipu

GLM-5.2

Rank 45 of 66 in Overall Models. All three benchmark settings tested.

Score · relative 0–100
44.5
Rank
45
Status
Complete 3/3
Rank change
0
Players generated
2026-07-02 – 2026-09-03
Match evidence updated
2026-10-26
Results published
Compare in charts View player programs Back to leaderboard

Scores by reasoning setting

  • XHigh (Intensive group, tested and included)44.5
  • High (Balanced group, tested and included)47.8
  • Mediumnot included (Balanced group, tested, not included)47.8
  • Highnot included (Balanced group, tested, not included)51.6
  • Mediumnot included (Balanced group, tested, not included)44.0
  • None (Baseline group, tested and included)41.3
  • Reasoningnot included (Baseline group, tested, not included)44.8
  • Reasoningnot included (Baseline group, tested, not included)34.3

Settings marked “not included” were tested but are left out of the overall score.

Available settings and grouping

Available settings

Reasoning disabledHighXHigh

Checked on 2026-07-21 · Provider documentation

How settings are grouped

  • Baseline: Reasoning disabled — tested and included
  • Balanced: High — tested and included
  • Intensive: XHigh — tested and included

Scores by game

  • Go40.6
  • Chess50.7
  • Shogi36.0
  • Xiangqi50.0
  • Hex49.4
  • International Draughts55.9
  • Game of the Amazons39.0
  • Tumbleweed34.6

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: High (47.8).

Across included settings, the highest game score is International Draughts (55.9); the lowest is Tumbleweed (34.6). 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
z-ai/glm-5.2
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 32 indexed attempts)

Generation evidence

  • Reasoning disabled: 16 contributing player programs across 8 games
  • High: 8 contributing player programs across 8 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 attempt54.11507.9176 / 184 / 156
GoBalanced2 working of 2 attempts39.1 program scores 46.0 · 32.21363.2 program range 1295–1431244 / 280 / 508
GoBaseline3 working of 3 attempts28.5 program scores 31.6 · 30.3 · 23.61254.1 program range 1198–1289265 / 301 / 982
ChessIntensive1 working of 1 attempt62.11569.3170 / 240 / 70
ChessBalanced2 working of 2 attempts48.9 program scores 53.9 · 43.91463.4 program range 1424–150295 / 716 / 149
ChessBaseline3 working of 3 attempts40.9 program scores 42.3 · 41.4 · 39.11401.6 program range 1387–141212 / 1108 / 320
ShogiIntensive1 working of 1 attempt37.31316.879 / 175 / 240
ShogiBalanced2 working of 2 attempts30.5 program scores 33.5 · 27.61254.6 program range 1227–128262 / 416 / 510
ShogiBaseline3 working of 3 attempts40.1 program scores 47.2 · 37.8 · 35.41341.5 program range 1300–1403290 / 501 / 691
XiangqiIntensive1 working of 1 attempt58.61543.0163 / 240 / 95
XiangqiBalanced2 working of 2 attempts48.3 program scores 56.4 · 40.21461.9 program range 1399–1525180 / 568 / 248
XiangqiBaseline3 working of 3 attempts43.1 program scores 43.8 · 42.9 · 42.61421.8 program range 1418–1427178 / 883 / 433
HexIntensive1 working of 1 attempt58.91550.2303 / 0 / 227
HexBalanced2 working of 2 attempts53.9 program scores 55.3 · 52.61486.6 program range 1470–1503554 / 0 / 506
HexBaseline3 working of 3 attempts35.5 program scores 59.8 · 27.0 · 19.71231.6 program range 1004–1563544 / 0 / 1046
International DraughtsIntensive1 working of 1 attempt42.41303.0177 / 26 / 299
International DraughtsBalanced2 working of 2 attempts78.7 program scores 85.8 · 71.61926.0 program range 1783–2069748 / 40 / 216
International DraughtsBaseline3 working of 3 attempts46.6 program scores 69.6 · 39.1 · 31.01377.5 program range 1132–1747619 / 48 / 839
Game of the AmazonsIntensive1 working of 1 attempt10.3843.047 / 0 / 483
Game of the AmazonsBalanced2 working of 2 attempts58.8 program scores 66.2 · 51.51563.7 program range 1466–1661602 / 0 / 458
Game of the AmazonsBaseline3 working of 3 attempts47.7 program scores 60.5 · 43.0 · 39.61420.1 program range 1318–1583720 / 0 / 870
TumbleweedIntensive1 working of 1 attempt31.91191.527 / 263 / 248
TumbleweedBalanced2 working of 2 attempts24.0 program scores 24.4 · 23.61102.3 program range 1098–1107109 / 256 / 711
TumbleweedBaseline3 working of 3 attempts47.9 program scores 82.3 · 31.3 · 30.01434.2 program range 1170–1948559 / 354 / 701

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

0 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: 88% / 12% / 0% · n 8High: 38% / 62% / 0% · n 16Reasoning disabled: 54% / 46% / 0% · n 24 All three benchmark settings testedMany programs needed repair

Generation cost

$0.10average estimated cost per game and reasoning setting

Median
$0.06
Range
$0.01–$0.39
Known cost subtotal (incomplete or unverified charges)
$3.51
Cost data
24 combinations · 48 runs
Price source
Provider-reported
Price list
Not recorded for this older data
Average output
35k tokens
Output range
1.4k–130k 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