DuelLabGameBench 2
Menu

Games · GB2-01

Go

Go is a large-grid territory-control game with stone placement, captures, ko restrictions, passing, and final area scoring. Scores compare AI-written player programs on DuelLab's rules for Go; they do not measure human play strength.

Updated
2026-10-22
Reasoning variants
229
Player programs
255

Leaderboard

By model and setting

Each model and reasoning setting appears once. If a model produced more than one player program for this game, DuelLab averages them before ranking. Score is relative within this game (0–100). Scores here are working-program strength: failed generation attempts are left out and count as 0 only in a model's headline score.

GB2-01: Go - Reasoning Variants
Rank Model Reasoning Score Raw rating W / D / L
1GPT-5.4Medium97.82324.8475 / 6 / 7
2GPT-6 AstraXHigh97.02259.2473 / 0 / 15
3Gemini 4 ArgonHigh95.52173.1242 / 0 / 16
4GPT-5.4XHigh94.92147.1455 / 13 / 20
5Claude Opus 5.5Medium93.52089.8439 / 32 / 17
6GPT-6 AstraLow92.92068.1439 / 24 / 25
7Claude Fable 5.1XHigh92.62060.5453 / 2 / 35
8Gemini 4 ArgonMedium92.52057.4227 / 0 / 25
9Pareto 26.10Provider92.42053.9257 / 0 / 23
10Gemini 4 ArgonLow91.62028.7232 / 0 / 22
11Claude Opus 5.5Low90.92008.5432 / 21 / 35
12Muse Spark OpenXHigh90.51997.9230 / 0 / 26
13Gemini 3.5 FlashHigh90.31994.8427 / 21 / 40
14Claude Fable 5.1Low88.91959.5427 / 4 / 58
15Claude Fable 5XHigh86.31906.6414 / 1 / 73
16Muse Spark 1.2Medium85.61892.6402 / 22 / 64
17Muse WonderXHigh84.91880.5406 / 5 / 77
18Grok 4.5High84.11865.9357 / 98 / 33
19Grok 4.5Medium84.11865.8384 / 41 / 63
20Claude Opus 5.5XHigh84.11865.6402 / 8 / 78
21GlimpseEnabled82.91844.8201 / 16 / 43
22DeepSeek V4.1 Flash 1017High82.51838.6212 / 0 / 50
23GPT-5.6 LunaXHigh81.71826.9392 / 1 / 95
24Muse Spark 1.2XHigh81.51823.4389 / 2 / 97
25GPT-5.4None81.01816.1365 / 46 / 77

B1 ratings carry no uncertainty term, so no rating-evidence index is shown; see Methodology.

Underlying results

Individual player programs

Each row is one accepted player program for GB2-01: Go; a model can have more than one. Score is the 0–100 result in this game, and its rating and W / D / L record show how it played. Scores are working-program strength: failed generation attempts have no row and count only in a model's headline score.

255 player programs.

ModelReasoningStatusScoreRatingW / D / LMatchesOpponentsNotesGenerated
GPT-5.4E-d4fa7dbb9458c464MediumRated Interval validated97.82325475 / 6 / 7488 / 4882446 capped draws2026-08-09First try
GPT-6 AstraE-fc5055140bf78475XHighRated Interval validated97.02259473 / 0 / 15488 / 488244—2026-09-05First try
Gemini 4 ArgonE-c9b7bf650530d8baHighRated Interval validated95.52173242 / 0 / 16258 / 258129—2026-10-21First try
GPT-5.4E-9879a37f54e31cffXHighRated Interval validated94.92147455 / 13 / 20488 / 48824413 capped draws2026-08-09First try
Gemini 3.6 FlashE-7484f4e7d8863a3bMediumRated Interval validated93.92104452 / 9 / 27488 / 4882449 capped draws2026-07-24First try
Claude Opus 5.5E-9e6850818d33c592MediumRated Interval validated93.52090439 / 32 / 17488 / 48824432 capped draws2026-09-22First try
GPT-6 AstraE-7fbf151da7860da4LowRated Interval validated92.92068439 / 24 / 25488 / 48824424 capped draws2026-09-05First try
Claude Fable 5.1E-443622d1ef9c6f19XHighRated Preliminary rev 976df3ba92.62061453 / 2 / 35490 / 4902452 capped draws2026-09-08First try
Gemini 4 ArgonE-31ef2459454f5e17MediumRated Interval validated92.52057227 / 0 / 25252 / 252126—2026-10-21First try
Pareto 26.10E-74fb208ee7158901ProviderRated Interval validated92.42054257 / 0 / 23280 / 280140—2026-10-21First try
Gemini 4 ArgonE-b29fd1d6479f1bc8LowRated Interval validated91.62029232 / 0 / 22254 / 254127—2026-10-21First try
Claude Opus 5E-700f3dddd611df07XHighRated Interval validated91.52026442 / 5 / 41488 / 4882445 capped draws2026-07-25First try
Claude Opus 5.5E-c6c2699b24183df3LowRated Interval validated90.92009432 / 21 / 35488 / 48824421 capped draws2026-09-22First try
Muse Spark OpenE-482f21724d9e9898XHighRated Interval validated90.51998230 / 0 / 26256 / 256128—2026-10-20First try
Gemini 3.5 FlashE-1802f43944b4b6f0HighRated Interval validated90.31995427 / 21 / 40488 / 48824421 capped draws2026-09-03First try
Claude Opus 5E-865d68ab219fd7eeNoneRated Interval validated89.61977404 / 63 / 23490 / 49024563 capped draws2026-07-25First try
Claude Fable 5.1E-d04c4d8635b641f9LowRated88.91959427 / 4 / 58489 / 4902454 capped draws2026-09-03First try
Claude Fable 5E-99c76fc062487714XHighRated Interval validated86.31907414 / 1 / 73488 / 4882441 capped draw2026-09-03First try
Muse Spark 1.2E-052cdf1e5975b860MediumRated Interval validated85.61893402 / 22 / 64488 / 48824422 capped draws2026-08-13First try
Muse WonderE-5c48ba07caee5554XHighRated Interval validated84.91881406 / 5 / 77488 / 4882445 capped draws2026-10-12First try
Grok 4.5E-1c8e452ede41ea2eHighRated Interval validated84.11866357 / 98 / 33488 / 48824498 capped draws2026-07-24Repaired
Grok 4.5E-d709aeb6802f9289MediumRated Interval validated84.11866384 / 41 / 63488 / 48824441 capped draws2026-07-24First try
Claude Opus 5.5E-05f63acd714d31e5XHighRated Interval validated84.11866402 / 8 / 78488 / 4882448 capped draws2026-09-22First try
GlimpseE-160753088cb3a492EnabledRated Interval validated82.91845201 / 16 / 43260 / 26013016 capped draws2026-10-21First try
DeepSeek V4.1 Flash 1017E-5074fe2d98bd9ce2HighRated Interval validated82.51839212 / 0 / 50262 / 262131—2026-10-20First try

Draws at the move limit count as draws; a crash or an illegal move counts as that program's loss and is also listed as a fault.