Name: DuelLab GameBench 2 public release gb2-20260807-76d592450aec
Creator: DuelLab
License: https://creativecommons.org/licenses/by/4.0/

Game profile

Shogi

Shogi is a 9x9 chess-family game with promotion zones, captured-piece drops, repetition handling, and impasse scoring.

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

By model and setting

GB2-03: Shogi by model and setting

Each model and reasoning setting appears once. If a model produced more than one accepted player program, DuelLab averages them before ranking. Raw rating, W / D / L, and uncertainty combine those contributing programs.

Reasoning level: Reasoning Variants Game: GB2-03: Shogi

GB2-03: Shogi - Reasoning Variants
Rank	Model	Reasoning	Score	Raw Elo	W / D / L	Uncertainty
1	Claude Opus 5	XHigh	98.4	1886.6	103 / 0 / 9	22.5
2	Claude Opus 5	Medium	95.7	1861.0	135 / 8 / 15	12.2
3	Claude Fable 5	XHigh	70.5	1700.3	27 / 30 / 7	18.3
4	GPT-5.6 Sol	XHigh	57.9	1616.4	35 / 7 / 26	16.4
5	Kimi K3	Low	57.2	1630.8	43 / 12 / 15	39.8
6	GPT-5.4	None	56.3	1602.3	31 / 32 / 17	11.8
7	Inkling	Medium	54.9	1607.2	24 / 8 / 14	29.2
8	Claude Opus 5	None	54.1	1586.1	84 / 34 / 54	9.9
9	GPT-5.6 Terra	XHigh	52.5	1584.7	27 / 15 / 16	21.3
10	Claude Sonnet 5	XHigh	52.2	1573.9	56 / 70 / 48	9.9
11	Gemini 3.5 Flash	Medium	51.1	1571.9	23 / 30 / 15	16.4
12	GPT-5.6 Sol	Medium	46.4	1545.0	29 / 13 / 16	21.3
13	GPT-5.5	XHigh	45.4	1530.4	21 / 30 / 29	11.8
14	Inkling	None	44.9	1526.9	49 / 78 / 37	11.2
15	GPT 4.1	None	44.7	1525.8	25 / 27 / 30	11.1
16	Gemini 3.6 Flash	High	44.7	1587.0	21 / 4 / 9	76.2
17	GPT-5.6 Terra	None	44.6	1525.7	20 / 35 / 23	12.5
18	Claude Opus 4.5	None	43.9	1518.8	13 / 60 / 15	9.2
19	Gemini 3.5 Flash Lite	Minimal	43.9	1538.2	33 / 27 / 24	33.5
20	Kimi K3	High	43.9	1565.5	19 / 18 / 9	67.8
21	Gemini 3.6 Flash	Minimal	43.3	1539.0	52 / 52 / 38	39.1
22	DeepSeek V4 Flash	Medium	43.0	1524.5	15 / 18 / 21	23.7
23	GPT-5.4 Nano	None	42.3	1508.4	28 / 32 / 28	9.2
24	GLM-5.2	XHigh	42.2	1533.6	27 / 21 / 18	42.0
25	Claude Opus 4.5	Historical	41.4	1510.3	19 / 26 / 17	19.2
26	Grok 4.5	Medium	41.3	1504.1	24 / 28 / 26	12.5
27	GPT-5.4 Mini	Medium	38.6	1485.6	15 / 48 / 20	10.8
28	GPT-5.6 Terra	Medium	38.2	1484.3	14 / 39 / 25	12.5
29	GLM-5.2	None	38.2	1492.0	23 / 57 / 32	22.5
30	Minimax M3	None	38.2	1480.9	32 / 33 / 25	8.7
31	GPT-5.4 Mini	None	37.9	1480.7	1 / 77 / 4	11.1
32	Gemini 3.5 Flash	High	37.5	1499.3	9 / 53 / 12	37.4
33	GPT-5.6 Luna	None	37.0	1476.9	21 / 37 / 18	13.2
34	Gemini 3.5 Flash Lite	High	35.8	1511.6	34 / 21 / 33	64.6
35	LongCat 2.0	Enabled	35.5	1502.1	16 / 22 / 12	57.1
36	MiMo-V2.5	Reasoning	35.3	1463.1	19 / 41 / 26	9.9
37	Gemini 3.6 Flash	Medium	34.8	1474.5	32 / 27 / 36	28.1
38	Laguna S 2.1	None	33.7	1454.9	39 / 71 / 46	12.7
39	GPT-5.5	None	33.6	1453.6	18 / 36 / 26	11.8
40	Hy3 Preview	None	33.3	1450.2	29 / 27 / 28	10.5
41	GPT-5.6 Sol	None	33.3	1451.2	15 / 41 / 24	11.8
42	MiMo-V2.5-Pro	None	33.0	1453.4	14 / 39 / 15	16.4
43	GPT-5.6 Luna	Medium	32.5	1449.2	11 / 42 / 17	15.6
44	Nex N2 Pro	None	32.5	1450.4	9 / 43 / 14	17.3
45	Claude Opus 4.8	Medium	32.3	1449.8	11 / 38 / 15	18.3
46	GPT-5.4	XHigh	31.9	1449.8	17 / 17 / 24	21.3
47	Laguna S 2.1	Enabled	31.9	1466.3	31 / 59 / 48	42.1
48	DeepSeek V4 Flash	None	31.6	1438.5	20 / 39 / 27	9.9
49	GPT-5.4	Medium	31.4	1439.4	18 / 36 / 24	12.5
50	GPT-5.5	Medium	31.4	1440.4	15 / 35 / 24	14.0
51	O3	High	31.2	1477.0	12 / 20 / 12	60.8
52	GLM-5.2	Medium	30.9	1446.0	3 / 38 / 11	24.9
53	Claude Opus 4.5	Medium	30.8	1438.8	12 / 34 / 22	16.4
54	Qwen3.7 Max	None	30.5	1437.8	6 / 50 / 10	17.3
55	Grok 4.5	High	30.0	1452.6	14 / 31 / 23	40.8
56	Grok Build 0.1	Enabled	29.9	1452.3	17 / 31 / 20	40.9
57	Inkling	Max	29.9	1489.6	9 / 13 / 12	76.2
58	Gemini 3.5 Flash Lite	Medium	29.2	1434.9	18 / 49 / 37	24.9
59	GPT-OSS 120B	Medium	29.2	1430.3	8 / 43 / 11	19.2
60	Kimi K3	Max	28.9	1492.4	7 / 14 / 11	76.2
61	Nex N2 Pro	Enabled	27.2	1432.5	13 / 33 / 26	38.4
62	LongCat 2.0	None	27.1	1410.8	37 / 74 / 51	11.5
63	GPT 5	High	26.8	1431.5	16 / 26 / 30	39.9
64	Claude Opus 4.5	High	26.7	1439.4	13 / 14 / 27	51.2
65	Nemotron 3 Ultra 550B A55B	None	25.7	1414.2	3 / 36 / 9	27.7
66	DeepSeek V4 Pro	None	25.5	1406.3	15 / 18 / 29	19.2
67	GPT 5	Medium	24.5	1399.4	8 / 35 / 19	19.2
68	Ling 3.0 Flash	None	23.7	1384.0	35 / 84 / 77	6.5
69	North Mini Code	Enabled	23.4	1414.2	9 / 27 / 24	46.2
70	GPT-OSS 120B	Low	21.2	1400.7	6 / 35 / 23	48.1
71	GPT-5.6 Luna	XHigh	15.8	1341.7	6 / 27 / 31	18.3
72	Grok 4.5	Low	6.7	1297.9	5 / 17 / 52	37.6

Underlying results

Individual player programs

Each row is one accepted player program for GB2-03: Shogi. 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 Reasoning Status Sort Showing 88 player programs.

Player ID	Model	Reasoning	Status	Score	Elo	Confidence	W / D / L	Rated / attempted	Opponents	Capped draws	Faults / players	Generation
E-04fa5b762ca703f0	Claude Fable 5	XHigh	Rated	70.5	1700	—	27 / 30 / 7	64 / 64	29	0	0 / 1	First try · 2026-07-03
E-1975db22db58e9b0	Claude Opus 4.5	Thinking field omitted	Rated	43.9	1519	—	13 / 60 / 15	88 / 88	44	13	0 / 1	First try · 2026-07-02
E-cef4890e88806969	Claude Opus 4.5	Legacy_shared_thinking_budget	Waiting	—	—	—	0 / 0 / 0	0 / 0	0	0	0 / 1	First try · 2026-07-02
E-d5810442d7094396	Claude Opus 4.5	High effort + 32,000 thinking tokens	Limited data	17.7	1385	—	4 / 6 / 14	24 / 24	12	1	0 / 1	First try · —
E-dfb18dba94eee30b	Claude Opus 4.5	Medium effort + 16,000 thinking tokens	Rated	30.8	1439	—	12 / 34 / 22	68 / 68	34	10	0 / 1	First try · —
E-823b961281482cd1	Claude Opus 4.8	Medium	Rated	32.3	1450	—	11 / 38 / 15	64 / 64	32	7	0 / 1	First try · 2026-07-02
E-38759212c025e8eb	Claude Opus 5	XHigh	Rated	100.0	1897	—	52 / 0 / 4	56 / 56	24	0	0 / 1	First try · 2026-07-25
E-49aaba66f6b19080	Claude Opus 5	Thinking disabled	Rated	63.4	1648	—	42 / 11 / 31	84 / 84	42	0	0 / 1	First try · 2026-07-25
E-7caa6c5081d57676	Claude Opus 5	XHigh	Rated	96.8	1876	—	51 / 0 / 5	56 / 56	24	0	0 / 1	First try · 2026-07-25
E-bd02f4c2907d0a7c	Claude Opus 5	Medium	Rated	92.6	1841	—	64 / 4 / 10	78 / 78	39	0	0 / 1	First try · 2026-07-25
E-d8babc5495b1a188	Claude Opus 5	Medium	Rated	98.9	1881	—	71 / 4 / 5	80 / 80	40	0	0 / 1	First try · 2026-07-25
E-ecf847457e616425	Claude Opus 5	Thinking disabled	Rated	44.7	1524	—	42 / 23 / 23	88 / 88	44	1	0 / 1	First try · 2026-07-25
E-1c251b669239429c	Claude Sonnet 5	XHigh	Rated	55.1	1595	—	24 / 36 / 16	76 / 76	35	1	0 / 1	First try · 2026-07-19
E-d49e8df479ed0377	Claude Sonnet 5	XHigh	Rated	49.4	1553	—	32 / 34 / 32	98 / 98	45	3	0 / 1	First try · 2026-07-19
E-0f669ae82234cf76	North Mini Code	Enabled	Limited data	20.5	1393	—	3 / 16 / 13	32 / 32	16	1	0 / 1	Repaired · 2026-07-02
E-32e04fe5ad231857	DeepSeek V4 Flash	High	Rated	43.0	1524	—	15 / 18 / 21	54 / 54	27	1	0 / 1	First try · 2026-07-01
E-ae9d59ec3a8a1d63	DeepSeek V4 Flash	Reasoning disabled	Rated	31.6	1439	—	20 / 39 / 27	86 / 86	43	0	0 / 1	Repaired · 2026-07-02
E-f82a334a0fd1fb8e	DeepSeek V4 Pro	Reasoning disabled	Rated	25.5	1406	—	15 / 18 / 29	62 / 62	29	1	0 / 1	Repaired · 2026-07-02
E-46c3bb79de67d970	Gemini 3.5 Flash	High	Limited data	34.7	1480	—	6 / 23 / 9	38 / 38	19	1	0 / 1	Repaired · 2026-07-02
E-8ed795ea8fd1e0aa	Gemini 3.5 Flash	Medium	Rated	51.1	1572	—	23 / 30 / 15	68 / 68	34	0	0 / 1	Repaired · 2026-07-02
E-28274045fbc35ae9	Gemini 3.5 Flash Lite	Minimal	Limited data	44.6	1548	—	13 / 14 / 7	34 / 34	17	1	0 / 1	Repaired · 2026-07-24
E-7ed90f1b43d80cf9	Gemini 3.5 Flash Lite	High	Limited data	36.1	1505	—	10 / 3 / 11	24 / 24	12	0	0 / 1	First try · 2026-07-24
E-c7e8525629d5d658	Gemini 3.5 Flash Lite	Medium	Rated	31.5	1450	—	11 / 19 / 22	52 / 52	26	2	0 / 1	First try · 2026-07-24
E-d485651284072880	Gemini 3.5 Flash Lite	High	Limited data	29.8	1464	—	6 / 7 / 11	24 / 24	12	1	0 / 1	First try · 2026-07-24
E-d9a1854998f1794f	Gemini 3.5 Flash Lite	Medium	Rated	26.8	1419	—	7 / 30 / 15	52 / 52	26	6	0 / 1	First try · 2026-07-24
E-1d4a922c11c37a8d	Gemini 3.6 Flash	Medium	Rated	41.4	1516	—	20 / 14 / 16	50 / 50	25	1	0 / 1	First try · 2026-07-24
E-2e4f025f94dafe33	Gemini 3.6 Flash	Medium	Rated	28.3	1433	—	12 / 13 / 20	45 / 45	23	0	0 / 1	First try · 2026-07-24
E-6743f1c221046d65	Gemini 3.6 Flash	Minimal	Limited data	51.0	1590	—	18 / 8 / 8	34 / 34	17	2	0 / 1	First try · 2026-07-24
E-80b8c1a18d6dd238	Gemini 3.6 Flash	Minimal	Limited data	39.0	1512	—	11 / 13 / 10	34 / 34	17	2	0 / 1	First try · 2026-07-24
E-97580ecf1bd06d40	Gemini 3.6 Flash	High	Limited data	51.3	1602	—	15 / 3 / 8	26 / 26	13	0	0 / 1	First try · 2026-07-24
E-c7da9914f86ee846	GPT 4.1	None	Rated	44.7	1526	—	25 / 27 / 30	82 / 82	41	2	0 / 1	Repaired · 2026-07-02
E-c541923774f0cd7b	GPT 5	Low	Waiting	—	—	—	0 / 0 / 0	0 / 0	0	0	0 / 1	Repaired · 2026-07-21
E-cc568b7069d5a0fd	GPT 5	High	Limited data	20.8	1399	—	6 / 8 / 14	28 / 29	14	1	1 / 1	First try · 2026-07-02
E-e724ba162431d43a	GPT 5	Medium	Rated	24.5	1399	—	8 / 35 / 19	62 / 62	31	1	0 / 1	First try · 2026-07-02
E-496f83ad0bd69447	GPT-5.4	Medium	Rated	31.4	1439	—	18 / 36 / 24	78 / 78	39	2	0 / 1	First try · 2026-07-02
E-4b75acaf58f14113	GPT-5.4	XHigh	Rated	31.9	1450	—	17 / 17 / 24	58 / 60	29	7	2 / 1	First try · 2026-07-02
E-ebd456dc6b320c61	GPT-5.4	None	Rated	56.3	1602	—	31 / 32 / 17	80 / 80	40	4	0 / 1	First try · 2026-07-02
E-35117a2e624d7970	GPT-5.4 Mini	None	Rated	37.9	1481	—	1 / 77 / 4	82 / 82	41	0	0 / 1	First try · 2026-07-02
E-9e39bdd48b7328c7	GPT-5.4 Mini	Medium	Rated	38.6	1486	—	15 / 48 / 20	83 / 83	42	11	0 / 1	Repaired · 2026-07-02
E-f2a7b43bf689e68e	GPT-5.4 Nano	None	Rated	42.3	1508	—	28 / 32 / 28	88 / 88	44	0	0 / 1	First try · 2026-07-09
E-446036823932df16	GPT-5.5	None	Rated	33.6	1454	—	18 / 36 / 26	80 / 80	40	7	0 / 1	First try · 2026-07-02
E-c002b0ea38b845f8	GPT-5.5	Medium	Rated	31.4	1440	—	15 / 35 / 24	74 / 74	37	7	0 / 1	First try · 2026-07-02
E-e53a932f820d9724	GPT-5.5	XHigh	Rated	45.4	1530	—	21 / 30 / 29	80 / 81	40	3	0 / 1	First try · 2026-07-02
E-0300c20cb034e3a4	GPT-5.6 Luna	Medium	Rated	32.5	1449	—	11 / 42 / 17	70 / 70	35	13	0 / 1	First try · 2026-07-10
E-3bca3b4e02f2acf9	GPT-5.6 Luna	XHigh	Rated	15.8	1342	—	6 / 27 / 31	64 / 64	31	2	0 / 1	First try · 2026-07-10
E-d192bb239087b778	GPT-5.6 Luna	None	Rated	37.0	1477	—	21 / 37 / 18	76 / 76	37	1	0 / 1	First try · 2026-07-10
E-2ac4e83a1b9ba989	GPT-5.6 Sol	XHigh	Rated	57.9	1616	—	35 / 7 / 26	68 / 68	29	0	0 / 1	First try · 2026-07-10
E-527999604d620080	GPT-5.6 Sol	None	Rated	33.3	1451	—	15 / 41 / 24	80 / 80	40	14	0 / 1	First try · 2026-07-10
E-b741de451c385e8e	GPT-5.6 Sol	Medium	Rated	46.4	1545	—	29 / 13 / 16	58 / 58	29	2	0 / 1	First try · 2026-07-10
E-39e763b7d49209b4	GPT-5.6 Terra	XHigh	Rated	52.5	1585	—	27 / 15 / 16	58 / 58	27	0	0 / 1	Repaired · 2026-07-10
E-c2f0f8f062659e83	GPT-5.6 Terra	None	Rated	44.6	1526	—	20 / 35 / 23	78 / 78	39	8	0 / 1	First try · 2026-07-10
E-e44c62bb88a3042a	GPT-5.6 Terra	Medium	Rated	38.2	1484	—	14 / 39 / 25	78 / 78	39	0	0 / 1	First try · 2026-07-10
E-2524de3c7d4a81b7	Ling 3.0 Flash	Baseline track (legacy evidence incomplete)	Rated	15.1	1328	—	14 / 39 / 47	100 / 100	41	1	0 / 1	Repaired · 2026-07-24
E-641ddca04b054d56	Ling 3.0 Flash	Baseline track (legacy evidence incomplete)	Rated	32.2	1440	—	21 / 45 / 30	96 / 96	39	0	0 / 1	Repaired · 2026-07-24
E-198e8ddfe9f7ce20	LongCat 2.0	None	Rated	17.5	1349	—	11 / 35 / 32	78 / 78	37	5	0 / 1	Repaired · 2026-07-24
E-84709c769628f9a2	LongCat 2.0	None	Rated	36.7	1473	—	26 / 39 / 19	84 / 84	40	9	0 / 1	First try · 2026-07-24
E-d3a5befe76215b48	LongCat 2.0	Enabled	Limited data	40.5	1524	—	12 / 13 / 7	32 / 32	16	2	0 / 1	First try · 2026-07-24
E-44cc758f104a74ab	Minimax M3	Reasoning disabled	Rated	38.2	1481	—	32 / 33 / 25	90 / 90	45	1	0 / 1	First try · 2026-07-02
E-09859be1b47131a0	Kimi K3	Low	Limited data	69.8	1710	—	28 / 7 / 3	38 / 38	19	0	0 / 1	First try · 2026-07-24
E-8f139a9e4d538859	Kimi K3	Max	Limited data	35.5	1498	—	6 / 11 / 9	26 / 26	13	0	0 / 1	First try · 2026-07-24
E-980b1f419104e27a	Kimi K3	High	Limited data	52.1	1598	—	14 / 14 / 6	34 / 34	17	1	0 / 1	First try · 2026-07-24
E-57ad2863cbf32d58	Nex N2 Pro	Reasoning enabled	Limited data	27.8	1437	—	5 / 18 / 13	36 / 36	18	2	0 / 1	Repaired · 2026-07-02
E-a05d4cef5f788f16	Nex N2 Pro	Reasoning disabled	Rated	32.5	1450	—	9 / 43 / 14	66 / 66	31	7	0 / 1	Repaired · 2026-07-02
E-eb6e8a61f620acfd	Nemotron 3 Ultra 550B A55B	Reasoning disabled	Rated	25.7	1414	—	3 / 36 / 9	48 / 48	24	6	0 / 1	Repaired · 2026-07-02
E-9f7b8b1adc2ddba7	O3	High	Limited data	29.1	1459	—	7 / 9 / 8	24 / 26	12	1	0 / 1	First try · 2026-07-03
E-87b4bd61da369b42	GPT-OSS 120B	Medium	Rated	29.2	1430	—	8 / 43 / 11	62 / 62	31	8	0 / 1	Repaired · 2026-07-02
E-c90a449ff5e7b79a	GPT-OSS 120B	Low	Rated	20.2	1381	—	3 / 28 / 13	44 / 44	22	2	0 / 1	First try · 2026-07-21
E-08762a3df1a52f6b	Laguna S 2.1	Enabled	Limited data	30.0	1462	—	3 / 18 / 5	26 / 26	13	1	0 / 1	First try · 2026-07-24
E-7acc466aeae25d41	Laguna S 2.1	None	Rated	34.7	1464	—	19 / 33 / 18	70 / 70	35	1	0 / 1	First try · 2026-07-24
E-91efdb0c216f8504	Laguna S 2.1	None	Rated	32.7	1446	—	20 / 38 / 28	86 / 86	43	5	0 / 1	Repaired · 2026-07-24
E-936fb7956df3d03b	Laguna S 2.1	Enabled	Limited data	36.8	1507	—	9 / 6 / 11	26 / 26	13	2	0 / 1	Repaired · 2026-07-24
E-3325ebad2ef000c1	Qwen3.7 Max	Reasoning disabled	Rated	30.5	1438	—	6 / 50 / 10	66 / 66	31	6	0 / 1	First try · 2026-07-02
E-2346cd049910cc0a	Hy3 Preview	None	Rated	33.3	1450	—	29 / 27 / 28	84 / 84	41	0	0 / 1	First try · 2026-07-02
E-77829668b3641188	Inkling	Medium	Rated	54.9	1607	—	24 / 8 / 14	46 / 46	23	2	0 / 1	First try · 2026-07-24
E-78048b7c08122f1c	Inkling	None	Rated	33.0	1450	—	18 / 42 / 18	78 / 78	37	6	0 / 1	Repaired · 2026-07-24
E-adf1f0bb193bd39b	Inkling	None	Rated	56.8	1604	—	31 / 36 / 19	86 / 86	43	9	0 / 1	First try · 2026-07-24
E-f9259c1ced4fba75	Inkling	Max	Limited data	32.2	1476	—	6 / 11 / 9	26 / 26	13	0	0 / 1	First try · 2026-07-24
E-02f7e54245bd5c85	Grok 4.5	Low	Rated	13.5	1340	—	3 / 11 / 26	40 / 40	20	0	0 / 1	Repaired · 2026-07-24
E-c0c60ca46f3ce498	Grok 4.5	High	Limited data	26.8	1432	—	5 / 18 / 11	34 / 34	17	3	0 / 1	First try · 2026-07-24
E-cbed46472ce4dd05	Grok 4.5	Medium	Rated	41.3	1504	—	24 / 28 / 26	78 / 78	39	1	0 / 1	First try · 2026-07-24
E-2a93343a8a88fd59	Grok Build 0.1	Reasoning enabled	Limited data	35.4	1487	—	8 / 19 / 9	36 / 36	18	0	0 / 1	Repaired · 2026-07-02
E-b10e12eb6aef112e	MiMo-V2.5	Reasoning disabled	Rated	35.3	1463	—	19 / 41 / 26	86 / 86	43	1	0 / 1	First try · 2026-07-02
E-b0948c64dddf57ad	MiMo-V2.5-Pro	Reasoning disabled	Rated	33.0	1453	—	14 / 39 / 15	68 / 68	32	5	0 / 1	Repaired · 2026-07-02
E-40aa28533b1afda1	GLM-5.2	High	Rated	30.9	1446	—	3 / 38 / 11	52 / 52	26	13	0 / 1	Repaired · 2026-07-11
E-4cab1ccd88bbeb2e	GLM-5.2	Reasoning disabled	Rated	34.8	1471	—	11 / 31 / 12	54 / 54	27	0	0 / 1	First try · 2026-07-02
E-583be34ca1fb2631	GLM-5.2	Reasoning disabled	Rated	41.6	1513	—	12 / 26 / 20	58 / 58	29	4	0 / 1	Repaired · 2026-07-11
E-d047860be3ae0d90	GLM-5.2	XHigh	Limited data	38.5	1509	—	14 / 8 / 12	34 / 34	17	3	0 / 1	First try · 2026-07-19
E-d3a5bd501d8146e0	GLM-5.2	XHigh	Limited data	45.8	1558	—	13 / 13 / 6	32 / 32	16	1	0 / 1	First try · 2026-07-19