Name: DuelLab GameBench 2 public release gb2-20260824-e05a1ff89a26
Creator: DuelLab
License: https://creativecommons.org/licenses/by/4.0/

Game profile

Tumbleweed

Tumbleweed is a hex-grid territory game where players grow visible stacks and final control is determined by line-of-sight strength.

Scores here compare player programs using DuelLab's encoded rules for Tumbleweed; they are not a broad measure of human play strength.

By model and setting

GB2-08: Tumbleweed by model and setting

Player programs are ranked by relative per-game score (0–100). Raw rating is shown as an advanced metric, alongside wins, draws, losses, and a per-game uncertainty index.

Reasoning level: Balanced Game: GB2-08: Tumbleweed

GB2-08: Tumbleweed - Balanced filter
Rank	Model	Score	Contributing programs	Raw Elo	W / D / L	Uncertainty
1	Grok 4.5	87.2	1	1917.1	80 / 1 / 13	7.5
2	Claude Opus 4.5	84.3	1	1878.7	140 / 2 / 26	0.0
3	Claude Opus 5	84.1	2	1888.4	156 / 0 / 26	8.4
4	Claude Fable 5	74.2	1	1794.8	72 / 0 / 12	10.5
5	Gemini 3.6 Flash	71.2	1	1755.9	88 / 21 / 33	0.0
6	Muse Spark 1.2	64.1	1	1696.9	55 / 16 / 19	8.7
7	Kimi K2.7 Code	63.1	1	1688.7	49 / 8 / 27	10.5
8	GPT-5.6 Sol	59.4	1	1652.2	53 / 20 / 17	8.7
9	GPT-5.5	55.4	1	1612.4	35 / 42 / 21	6.5
10	Qwen3.8 Max	52.2	1	1632.7	11 / 5 / 2	71.0
11	GPT-5.4 Nano	51.1	2	1574.6	67 / 46 / 53	10.8
12	GPT-5.4	41.5	1	1478.0	27 / 53 / 23	5.3
13	O3	41.0	1	1471.6	28 / 57 / 27	3.3
14	GPT-5.6 Terra	40.7	1	1472.2	18 / 48 / 30	7.0
15	Gemini 3.1 Flash Lite	37.0	1	1432.5	36 / 37 / 45	2.1
16	Ox Alpha	36.8	3	1497.4	12 / 17 / 13	85.3
17	GPT-5.4 Mini	36.4	1	1429.5	33 / 37 / 34	5.0
18	Qwen3.7 Max	36.3	1	1427.2	9 / 71 / 28	4.1
19	GPT-OSS 120B	36.2	1	1426.3	9 / 79 / 20	4.1
20	Qwen3.7 Plus	35.3	1	1417.7	0 / 88 / 18	4.6
21	Minimax M3	35.2	1	1417.7	7 / 81 / 16	5.0
22	Nemotron 3 Ultra 550B A55B	35.0	1	1417.7	9 / 65 / 22	7.0
23	Grok Build 0.1	34.6	1	1412.9	6 / 81 / 13	6.0
24	GLM-5.2	34.6	1	1422.0	1 / 55 / 10	17.3
25	Claude Opus 4.8	34.1	1	1408.5	5 / 75 / 18	6.5
26	DeepSeek V4 Flash	33.5	2	1409.5	4 / 128 / 12	14.8
27	Inkling Small	33.3	1	1460.1	0 / 12 / 3	81.2
28	Kimi K3	32.3	1	1380.0	12 / 129 / 43	0.0
29	MiMo-V2.5	31.6	2	1387.0	22 / 84 / 66	9.9
30	MiMo-V2.5-Pro	31.4	1	1385.0	2 / 66 / 20	9.2
31	Claude Sonnet 5	31.3	1	1384.2	7 / 52 / 29	9.2
32	LongCat 2.0	31.0	2	1370.9	60 / 95 / 150	0.0
33	Laguna S 2.1	29.6	2	1355.6	14 / 224 / 94	0.0
34	GPT 5	28.8	1	1362.3	7 / 51 / 20	12.5
35	Gemini 3.5 Flash	28.6	1	1360.7	0 / 60 / 18	12.5
36	North Mini Code	27.9	2	1370.0	0 / 71 / 19	33.2
37	Gemini 3.5 Flash Lite	27.2	2	1334.2	13 / 209 / 90	0.0
38	Inkling	26.6	1	1329.0	11 / 92 / 50	0.0
39	GPT-5.6 Luna	25.1	1	1323.7	3 / 55 / 32	8.7
40	Nex N2 Pro	13.7	1	1217.2	7 / 13 / 60	11.8

Underlying results

Individual player programs

Each row is one accepted player program for GB2-08: Tumbleweed. A model can appear more than once when it produced more than one program. Score is the normalized 0–100 result; Elo and W / D / L show the underlying play record. Confidence and evidence can be blank when too few compatible matches are available. Draws at the move limit count; one-sided model faults are listed separately.

Search model Status Sort Showing 50 player programs.

Player ID	Model	Status	Score	Elo	Confidence	W / D / L	Rated / attempted	Opponents	Capped draws	Faults / players	Generation
E-9ccf296d44d35675	Claude Fable 5	Rated	74.2	1795	—	72 / 0 / 12	84 / 84	41	0	0 / 1	First try · 2026-07-02
E-355427eb0d526f34	Claude Opus 4.5	Waiting	—	—	—	0 / 0 / 0	0 / 0	0	0	0 / 1	First try · 2026-07-02
E-72233463ddce5549	Claude Opus 4.5	Rated	84.3	1879	—	140 / 2 / 26	168 / 168	69	2	0 / 1	First try · 2026-07-21
E-c2dade0b97891d8a	Claude Opus 4.8	Rated	34.1	1409	—	5 / 75 / 18	98 / 98	48	75	0 / 1	First try · 2026-07-02
E-35a5bf9b35d5e1e8	Claude Opus 5	Rated	76.2	1813	—	72 / 0 / 18	90 / 90	45	0	0 / 1	First try · 2026-07-25
E-e73b6fb78880341d	Claude Opus 5	Rated	92.0	1964	—	84 / 0 / 8	92 / 92	46	0	0 / 1	First try · 2026-07-25
E-01c56a1fc7407264	Claude Sonnet 5	Rated	31.3	1384	—	7 / 52 / 29	88 / 88	42	52	0 / 1	First try · 2026-08-09
E-efd7b75d5cd3b538	North Mini Code	Rated	28.8	1368	—	0 / 44 / 16	60 / 60	30	44	0 / 1	First try · 2026-07-02
E-43aba68125a9abc5	DeepSeek V4 Flash	Rated	35.8	1431	—	2 / 63 / 7	72 / 72	36	63	0 / 1	First try · 2026-07-02
E-d28ef17ff6995c62	DeepSeek V4 Flash	Rated	31.3	1388	—	2 / 65 / 5	72 / 72	36	65	0 / 1	Repaired · 2026-07-01
E-cc6fb78c96472032	Gemini 3.1 Flash Lite	Rated	37.0	1432	—	36 / 37 / 45	118 / 118	59	37	0 / 1	First try · 2026-07-02
E-bd0bbefbbd252fc5	Gemini 3.5 Flash	Rated	28.6	1361	—	0 / 60 / 18	78 / 78	39	60	0 / 1	Repaired · 2026-07-02
E-2f2aa5b1beafd6e6	Gemini 3.5 Flash Lite	Rated	26.8	1330	—	7 / 109 / 44	160 / 160	60	109	0 / 1	First try · 2026-07-24
E-9b50de854a87d3c6	Gemini 3.5 Flash Lite	Rated	27.6	1339	—	6 / 100 / 46	152 / 152	56	100	0 / 1	First try · 2026-07-24
E-ab7ee826f8050f04	Gemini 3.6 Flash	Rated	71.2	1756	—	88 / 21 / 33	142 / 142	59	21	0 / 1	First try · 2026-07-24
E-04ae9f2b42e0e036	GPT 5	Rated	28.8	1362	—	7 / 51 / 20	78 / 78	37	51	0 / 1	First try · 2026-07-02
E-6b601c15cf117b79	GPT-5.4	Rated	41.5	1478	—	27 / 53 / 23	103 / 103	47	53	0 / 1	First try · 2026-08-09
E-b5171a15f0048555	GPT-5.4 Mini	Rated	36.4	1429	—	33 / 37 / 34	104 / 104	49	37	0 / 1	First try · 2026-07-02
E-26fa12e871bd8f3a	GPT-5.4 Nano	Rated	25.5	1330	—	4 / 46 / 32	82 / 82	40	46	0 / 1	Repaired · 2026-07-02
E-e361d8cbba2abd67	GPT-5.4 Nano	Rated	76.8	1819	—	63 / 0 / 21	84 / 84	41	0	0 / 1	Repaired · 2026-07-09
E-0d83a3c5c5e66085	GPT-5.5	Rated	55.4	1612	—	35 / 42 / 21	98 / 98	48	42	0 / 1	First try · 2026-07-02
E-283f17e4d25ffd65	GPT-5.6 Luna	Rated	25.1	1324	—	3 / 55 / 32	90 / 90	42	55	0 / 1	First try · 2026-08-09
E-55079a10c13f4211	GPT-5.6 Sol	Rated	59.4	1652	—	53 / 20 / 17	90 / 90	42	20	0 / 1	First try · 2026-08-09
E-d004e78844cdd202	GPT-5.6 Terra	Rated	40.7	1472	—	18 / 48 / 30	96 / 96	47	48	0 / 1	First try · 2026-07-10
E-d724da56cbc06da7	LongCat 2.0	Rated	5.3	1124	—	1 / 46 / 117	164 / 164	59	46	0 / 1	First try · 2026-07-24
E-ef389910f6cc07cc	LongCat 2.0	Rated	56.7	1618	—	59 / 49 / 33	141 / 141	58	49	0 / 1	First try · 2026-07-24
E-502a038b3a2d63ef	Muse Spark 1.2	Rated	64.1	1697	—	55 / 16 / 19	90 / 90	42	16	0 / 1	First try · 2026-08-13
E-d169e7b29be70652	Minimax M3	Rated	35.2	1418	—	7 / 81 / 16	104 / 104	47	81	0 / 1	Repaired · 2026-07-02
E-251b5d53a5f89562	Kimi K2.7 Code	Rated	63.1	1689	—	49 / 8 / 27	84 / 84	39	8	0 / 1	Repaired · 2026-07-02
E-151c4ee6a2d1838e	Kimi K3	Rated	32.3	1380	—	12 / 129 / 43	184 / 184	71	129	0 / 1	First try · 2026-07-24
E-92332f4d765b2297	Nex N2 Pro	Rated	13.7	1217	—	7 / 13 / 60	80 / 80	38	13	0 / 1	Repaired · 2026-07-02
E-15cf0dede27b7e1b	Nemotron 3 Ultra 550B A55B	Rated	35.0	1418	—	9 / 65 / 22	96 / 96	48	65	0 / 1	Repaired · 2026-07-02
E-35de956c8305ff90	O3	Rated	41.0	1472	—	28 / 57 / 27	112 / 112	55	57	0 / 1	First try · 2026-07-02
E-6e65dec497ff2de3	GPT-OSS 120B	Rated	36.2	1426	—	9 / 79 / 20	108 / 108	51	79	0 / 1	Repaired · 2026-07-02
E-0fab95afbec0dd98	Laguna S 2.1	Rated	29.1	1351	—	7 / 111 / 48	166 / 166	63	111	0 / 1	First try · 2026-07-24
E-f72c72b8f264c57a	Laguna S 2.1	Rated	30.0	1360	—	7 / 113 / 46	166 / 166	63	113	0 / 1	First try · 2026-07-24
E-64624689e7cbd439	Qwen3.7 Max	Rated	36.3	1427	—	9 / 71 / 28	108 / 108	50	71	0 / 1	Repaired · 2026-07-02
E-6c2a9942e92223c8	Qwen3.7 Plus	Rated	35.3	1418	—	0 / 88 / 18	106 / 106	50	88	0 / 1	First try · 2026-07-02
E-58e68aa5bf529b8a	Qwen3.8 Max	Limited data	52.2	1633	—	11 / 5 / 2	18 / 18	7	5	0 / 1	First try · 2026-08-08
E-273045858e4a7f82	Ox Alpha	Rated	35.5	1484	—	3 / 7 / 4	14 / 14	7	7	0 / 1	First try · 2026-08-22
E-8094ab91b5be5ad0	Ox Alpha	Rated	43.7	1563	—	8 / 3 / 3	14 / 14	7	3	0 / 1	First try · 2026-08-22
E-a87ab54ca8c5ca34	Ox Alpha	Rated	31.3	1444	—	1 / 7 / 6	14 / 14	7	7	0 / 1	Repaired · 2026-08-22
E-9f74dab09049a473	Inkling	Rated	26.6	1329	—	11 / 92 / 50	153 / 153	58	92	0 / 1	First try · 2026-07-24
E-9cd2fd1302bb65bd	Inkling Small	Limited data	33.3	1460	—	0 / 12 / 3	15 / 15	7	12	0 / 1	First try · 2026-08-09
E-73422318071a105c	Grok 4.5	Rated	87.2	1917	—	80 / 1 / 13	94 / 94	47	1	0 / 1	First try · 2026-07-24
E-3e2fe948d2bcebb0	Grok Build 0.1	Rated	34.6	1413	—	6 / 81 / 13	100 / 100	50	81	0 / 1	Repaired · 2026-07-02
E-650d06103e7fa176	MiMo-V2.5	Rated	43.1	1496	—	20 / 44 / 26	90 / 90	43	44	0 / 1	Repaired · 2026-07-03
E-b5eb7da0f5021b69	MiMo-V2.5	Rated	20.1	1278	—	2 / 40 / 40	82 / 82	38	40	0 / 1	Repaired · 2026-07-02
E-bd1bb7ac29877b8f	MiMo-V2.5-Pro	Rated	31.4	1385	—	2 / 66 / 20	88 / 88	41	66	0 / 1	Repaired · 2026-07-02
E-da1e9a6be31e371f	GLM-5.2	Rated	34.6	1422	—	1 / 55 / 10	66 / 66	33	55	0 / 1	Repaired · 2026-07-11