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

Game profile

Xiangqi

Xiangqi is a 9x10 chess-family game with palaces, river restrictions, cannon screens, horse-leg and elephant-eye blocking, and flying-general legality.

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

By model and setting

GB2-04: Xiangqi 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: Baseline Game: GB2-04: Xiangqi

GB2-04: Xiangqi - Baseline filter
Rank	Model	Score	Contributing programs	Raw Elo	W / D / L	Uncertainty
1	LongCat 2.0	55.2	1	1580.2	17 / 44 / 19	11.8
2	Claude Opus 5	51.3	2	1554.3	52 / 63 / 49	11.1
3	Claude Opus 4.5	48.9	1	1544.3	11 / 39 / 14	18.3
4	Kimi K3	47.2	1	1516.3	42 / 78 / 30	0.0
5	GPT-5.6 Terra	45.9	1	1522.6	7 / 60 / 5	14.8
6	Gemini 3.5 Flash Lite	45.7	2	1510.4	48 / 154 / 44	1.2
7	GPT-5.6 Sol	42.6	1	1497.9	13 / 54 / 15	11.1
8	Hy3 Preview	42.1	1	1503.3	12 / 36 / 10	21.3
9	MiMo-V2.5	42.0	2	1501.2	19 / 98 / 7	19.5
10	Claude Opus 4.8	41.8	1	1494.5	8 / 56 / 12	13.2
11	Claude Sonnet 5	41.6	1	1493.7	13 / 44 / 17	14.0
12	GPT-5.5	41.0	1	1487.3	10 / 58 / 16	10.5
13	GLM-5.2	40.7	2	1493.7	20 / 73 / 25	21.0
14	Ling 3.0 Flash	40.6	1	1480.7	7 / 90 / 5	5.5
15	Nex N2 Pro	40.6	1	1483.9	5 / 74 / 7	9.9
16	Muse Spark 1.2	40.5	1	1486.0	13 / 54 / 9	13.2
17	Gemini 3.6 Flash	40.3	2	1473.6	22 / 206 / 38	0.0
18	Gemma 4 31B	40.1	1	1482.3	10 / 61 / 9	11.8
19	Nemotron 3 Ultra 550B A55B	38.6	1	1484.4	5 / 39 / 6	26.3
20	Inkling Small	38.2	1	1525.9	2 / 13 / 0	81.2
21	GPT-5.4 Mini	38.0	1	1469.1	16 / 52 / 12	11.8
22	GPT-5.6 Luna	37.1	1	1463.5	11 / 52 / 15	12.5
23	Qwen3.7 Plus	36.9	1	1467.8	5 / 49 / 8	19.2
24	Inkling	36.2	2	1457.7	22 / 105 / 29	12.5
25	GPT 4.1	33.7	1	1443.1	4 / 56 / 14	14.0
26	Grok 4.5	33.6	1	1431.8	16 / 91 / 20	0.4
27	GPT-5.4	30.7	1	1420.9	12 / 42 / 29	10.8
28	Qwen3.8 Max	30.5	1	1459.6	1 / 17 / 4	60.5
29	Ox Alpha	26.9	3	1449.7	3 / 29 / 16	77.5
30	DeepSeek V4 Flash	22.5	1	1374.2	3 / 41 / 20	18.3
31	GPT-OSS 120B	0.0	1	1214.9	9 / 50 / 67	0.6

Underlying results

Individual player programs

Each row is one accepted player program for GB2-04: Xiangqi. 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 40 player programs.

Player ID	Model	Status	Score	Elo	Confidence	W / D / L	Rated / attempted	Opponents	Faults / players	Generation
E-9a77e43e752532c8	Claude Opus 4.5	Rated	48.9	1544	—	11 / 39 / 14	64 / 64	32	0 / 1	First try · 2026-07-02
E-688a139ef8557587	Claude Opus 4.8	Rated	41.8	1495	—	8 / 56 / 12	76 / 76	37	0 / 1	First try · 2026-07-02
E-72e9307528761b35	Claude Opus 5	Rated	40.8	1486	—	14 / 44 / 26	84 / 84	42	0 / 1	First try · 2026-07-25
E-9e7782014171d35f	Claude Opus 5	Rated	61.8	1623	—	38 / 19 / 23	80 / 80	40	0 / 1	First try · 2026-07-25
E-d4f6be4f3a2580ef	Claude Sonnet 5	Rated	41.6	1494	—	13 / 44 / 17	74 / 74	36	0 / 1	First try · 2026-08-09
E-81bc9a578754f220	DeepSeek V4 Flash	Rated	22.5	1374	—	3 / 41 / 20	64 / 64	32	0 / 1	First try · 2026-07-02
E-33e6ff93f425b731	Gemini 3.5 Flash	Waiting	—	—	—	0 / 0 / 0	0 / 0	0	0 / 1	First try · 2026-07-21
E-6a79049fd9ab6f5d	Gemini 3.5 Flash Lite	Rated	45.2	1507	—	20 / 82 / 20	122 / 128	48	0 / 1	First try · 2026-07-24
E-b05b0185d0e8dc2a	Gemini 3.5 Flash Lite	Rated	46.2	1513	—	28 / 72 / 24	124 / 124	47	0 / 1	Repaired · 2026-07-24
E-0b528ba96380a3b5	Gemini 3.6 Flash	Rated	39.2	1467	—	10 / 111 / 14	135 / 135	49	0 / 1	First try · 2026-07-24
E-846809faa1345207	Gemini 3.6 Flash	Rated	41.3	1481	—	12 / 95 / 24	131 / 131	50	0 / 1	Repaired · 2026-07-24
E-ec0cf4b37c805eee	Gemma 4 31B	Rated	40.1	1482	—	10 / 61 / 9	80 / 80	39	0 / 1	First try · 2026-07-02
E-ecb06954153be8d5	GPT 4.1	Rated	33.7	1443	—	4 / 56 / 14	74 / 74	37	0 / 1	First try · 2026-07-02
E-5cd71ee548acca75	GPT-5.4	Rated	30.7	1421	—	12 / 42 / 29	83 / 83	37	0 / 1	First try · 2026-08-09
E-0880cdf1a43cb90c	GPT-5.4 Mini	Rated	38.0	1469	—	16 / 52 / 12	80 / 80	39	0 / 1	First try · 2026-07-02
E-ba1aa87056709e9a	GPT-5.5	Rated	41.0	1487	—	10 / 58 / 16	84 / 84	41	0 / 1	First try · 2026-07-02
E-2c7615f28028f790	GPT-5.6 Luna	Rated	37.1	1463	—	11 / 52 / 15	78 / 78	38	0 / 1	First try · 2026-08-09
E-66db807b1be40f67	GPT-5.6 Sol	Rated	42.6	1498	—	13 / 54 / 15	82 / 82	37	0 / 1	First try · 2026-08-09
E-50ee0aadbc66fe0e	GPT-5.6 Terra	Rated	45.9	1523	—	7 / 60 / 5	72 / 72	36	0 / 1	First try · 2026-07-10
E-f38039bb8cc277a1	Ling 3.0 Flash	Rated	40.6	1481	—	7 / 90 / 5	102 / 102	40	0 / 1	First try · 2026-07-24
E-9e1d4bf411e97fa6	LongCat 2.0	Rated	55.2	1580	—	17 / 44 / 19	80 / 80	40	0 / 1	First try · 2026-07-24
E-3f1d300b580e2822	Muse Spark 1.2	Rated	40.5	1486	—	13 / 54 / 9	76 / 76	36	0 / 1	First try · 2026-08-13
E-3c06949f23d4a77f	Kimi K3	Rated	47.2	1516	—	42 / 78 / 30	150 / 150	60	0 / 1	Repaired · 2026-07-24
E-6f4b3b55e4ec7cf5	Nex N2 Pro	Rated	40.6	1484	—	5 / 74 / 7	86 / 86	42	0 / 1	Repaired · 2026-07-11
E-f36b99b3d76fbb35	Nemotron 3 Ultra 550B A55B	Rated	38.6	1484	—	5 / 39 / 6	50 / 50	25	0 / 1	First try · 2026-07-02
E-f6de1ce39e48807c	GPT-OSS 120B	Rated	0.0	1215	—	9 / 50 / 67	126 / 126	49	0 / 1	Repaired · 2026-07-21
E-84b9a07f75f324e2	Qwen3.7 Plus	Rated	36.9	1468	—	5 / 49 / 8	62 / 62	30	0 / 1	First try · 2026-07-02
E-91e0eadd75202409	Qwen3.8 Max	Limited data	30.5	1460	—	1 / 17 / 4	22 / 22	11	0 / 1	First try · 2026-08-08
E-2aaa1a2095c3d9b6	Ox Alpha	Rated	26.2	1445	—	0 / 11 / 5	16 / 16	8	0 / 1	First try · 2026-08-22
E-92dc825a33843ff8	Ox Alpha	Rated	23.9	1431	—	1 / 8 / 7	16 / 16	8	0 / 1	Repaired · 2026-08-22
E-ac806e30e1bd1958	Ox Alpha	Rated	30.5	1473	—	2 / 10 / 4	16 / 16	8	0 / 1	Repaired · 2026-08-22
E-3fd10f7865cc280e	Hy3 Preview	Rated	42.1	1503	—	12 / 36 / 10	58 / 58	29	0 / 1	Repaired · 2026-07-02
E-646ff33c7bab8b13	Inkling	Rated	32.3	1433	—	18 / 43 / 19	80 / 80	39	0 / 1	First try · 2026-07-24
E-ddc16fa197bfc7c5	Inkling	Rated	40.0	1483	—	4 / 62 / 10	76 / 76	37	0 / 1	First try · 2026-07-24
E-a50e6392704c24e8	Inkling Small	Limited data	38.2	1526	—	2 / 13 / 0	15 / 15	8	0 / 1	First try · 2026-08-09
E-5a4033db1131bf1d	Grok 4.5	Rated	33.6	1432	—	16 / 91 / 20	127 / 127	48	0 / 1	First try · 2026-07-24
E-7ad356726cce2b22	MiMo-V2.5	Rated	43.4	1507	—	16 / 48 / 4	68 / 68	34	0 / 1	Repaired · 2026-07-01
E-d58c55e107783a21	MiMo-V2.5	Rated	40.7	1495	—	3 / 50 / 3	56 / 56	28	0 / 1	Repaired · 2026-07-02
E-1d5f5ff0402ff944	GLM-5.2	Rated	46.8	1535	—	8 / 39 / 7	54 / 54	27	0 / 1	Repaired · 2026-07-02
E-f776c26a98306f5d	GLM-5.2	Rated	34.6	1452	—	12 / 34 / 18	64 / 64	32	0 / 1	First try · 2026-07-11