榜单

本站不自创综合总分；两榜不合并、不换算。

01 / Artificial Analysis

智能指数

最后更新 2026-11-08

数据说明

发布方: Artificial Analysis · 来源 ↗ · 本站不自创综合总分；两榜不合并、不换算。

照录 Artificial Analysis 免费端点 /api/v2/language/models/free（tier=free，指数版本 v5）的原样快照：共 726 条模型记录，其中 109 条带 Intelligence Index，按指数降序编号（同分同名次）。指数、中位输出速度、输入/输出价均为官方返回值，本站不改写、不加权、不自创总分。其中 83 条是官方按推理档位分列的同名条目，档位见「档位」列。未收录进本站的模型同样在榜，仅不可点击。

排名	模型	档位	厂商	指数 / 偏好分	中位速度	输入 / 输出价
1	Gemini 4 Carbon	High	Google	60	94.4 t/s	$4.00 / $20.00
2	Claude Opus 5.5	Max	Anthropic	59.1	95.1 t/s	$4.00 / $20.00
3	Gemini 4 Carbon	Medium	Google	58.3	93.2 t/s	$4.00 / $20.00
4	Claude Opus 5.5	Xhigh	Anthropic	57.4	81.9 t/s	$4.00 / $20.00
5	GPT-6 Astra	Max	OpenAI	56.9	50.3 t/s	$10.00 / $50.00
6	GPT-6 Astra	Xhigh	OpenAI	55.3	47.4 t/s	$10.00 / $50.00
7	GPT-6 Astra	High	OpenAI	54.8	49.3 t/s	$10.00 / $50.00
8	GPT-6.1 Sol	Max	OpenAI	54.6	68.7 t/s	$2.00 / $10.00
9	Claude Sonnet 5.5	Max	Anthropic	54.5	136.1 t/s	$2.00 / $10.00
10	Gemini 4 Carbon	Low	Google	54.2	83 t/s	$4.00 / $20.00
11	GPT-6.1 Sol	Xhigh	OpenAI	52.5	70.8 t/s	$2.00 / $10.00
12	Claude Opus 5.5	High	Anthropic	51.8	77.7 t/s	$4.00 / $20.00
13	GPT-6 Astra	Medium	OpenAI	51.6	43.4 t/s	$10.00 / $50.00
14	GPT-6.1 Terra	Max	OpenAI	51.3	108.6 t/s	$1.00 / $5.00
15	Gemini 4 Argon	High	Google	50.4	91 t/s	$4.00 / $20.00
16	GPT-6.1 Sol	High	OpenAI	50.3	63.6 t/s	$2.00 / $10.00
17	Muse Wonder Full	Max	Meta	50.1	—	—
18	Claude Fable 5.1	Xhigh	Anthropic	50	61.4 t/s	$10.00 / $50.00
19	Claude Sonnet 5.5	Xhigh	Anthropic	49.9	105.8 t/s	$2.00 / $10.00
20	Claude Fable 5.1	Max	Anthropic	49.2	69.4 t/s	$10.00 / $50.00
21	Claude Opus 5.5	Medium	Anthropic	48.2	76.7 t/s	$4.00 / $20.00
22	Claude Fable 5.1	High	Anthropic	47.2	56.7 t/s	$10.00 / $50.00
23	Kimi K3.1 Pro	—	Kimi	46.5	28.6 t/s	$2.00 / $10.00
24	GPT-6.1 Sol	Medium	OpenAI	46.3	58.4 t/s	$2.00 / $10.00
25	GPT-6.1 Terra	Xhigh	OpenAI	45.3	92 t/s	$1.00 / $5.00
26	Claude Opus 5	Max	Anthropic	44.2	50.6 t/s	$5.00 / $25.00
26	GPT-6 Sol	Max	OpenAI	44.2	142.2 t/s	$2.00 / $10.00
28	Muse Wonder	Max	Meta	43.6	143.7 t/s	$4.00 / $20.00
29	GPT-6 Astra	Low	OpenAI	43.3	42 t/s	$10.00 / $50.00
30	Claude Fable 5.1	Medium	Anthropic	43.2	57 t/s	$10.00 / $50.00
31	GPT-6.1 Terra	High	OpenAI	42.4	89.5 t/s	$1.00 / $5.00
32	Claude Sonnet 5.5	High	Anthropic	41.5	108.2 t/s	$2.00 / $10.00
33	Muse Wonder	High	Meta	40.5	149.4 t/s	$4.00 / $20.00
34	Claude Fable 5.1	Low	Anthropic	38.7	55.7 t/s	$10.00 / $50.00
35	GPT-6.1 Terra	Medium	OpenAI	38.2	89.8 t/s	$1.00 / $5.00
36	Muse Spark 1.3	Max	Meta	36.6	159.2 t/s	$1.25 / $4.25
37	Claude Haiku 5.5	Max	Anthropic	36.5	232.8 t/s	$0.10 / $0.50
38	Grok 4.7	Xhigh	SpaceXAI	35.5	65 t/s	$2.00 / $6.00
39	Grok 4.7	High	SpaceXAI	35.2	73.5 t/s	$2.00 / $6.00
40	Claude Opus 5.5	Low	Anthropic	34.8	77.3 t/s	$4.00 / $20.00
40	GPT-6.1 Sol	Low	OpenAI	34.8	59.8 t/s	$2.00 / $10.00
42	Claude Haiku 5.5	Xhigh	Anthropic	34.6	177.9 t/s	$0.10 / $0.50
42	Tencent Hy4	—	Tencent	34.6	93.8 t/s	$0.83 / $2.50
44	Qwen3.8 Max	0902	Alibaba	33.7	40 t/s	$2.00 / $6.00
45	DeepSeek V4.1 Flash 1017	Max	DeepSeek	32.6	147.7 t/s	$0.30 / $1.20
46	GPT-5.6 Terra	Max	OpenAI	32.3	120.7 t/s	$2.00 / $12.00
47	GLM-5.3	Max	Z AI	31.7	81 t/s	$1.40 / $4.40
48	Grok 4.7	Low	SpaceXAI	31.1	54.4 t/s	$2.00 / $6.00
49	Grok 4.6	Xhigh	SpaceXAI	31	72.1 t/s	$2.00 / $6.00
50	Claude Sonnet 5.5	Medium	Anthropic	30.8	103.4 t/s	$2.00 / $10.00
50	Gemini 3.8 Flash	High	Google	30.8	122.9 t/s	$0.75 / $3.75
52	Inkling 2 Small	—	Thinking Machines	30.6	128.3 t/s	$0.30 / $1.00
53	GPT-6.1 Terra	Non-reasoning	OpenAI	30.5	88.5 t/s	$1.00 / $5.00
54	Gemini 3.8 Flash	Medium	Google	30.1	126.8 t/s	$0.75 / $3.75
55	Claude Haiku 5.5	High	Anthropic	29.9	174.2 t/s	$0.10 / $0.50
55	MiMo-V2.6-Pro	—	Xiaomi	29.9	39.8 t/s	$0.43 / $0.87
57	Ling 3.1 Flash 1019	—	InclusionAI	29.6	—	$0.30 / $0.90
58	GPT-6.1 Terra	Low	OpenAI	29.4	89 t/s	$1.00 / $5.00
59	MiniMax-M3.1-Flash	—	MiniMax	29.1	166.7 t/s	$0.15 / $0.60
60	DeepSeek V4.1 Flash 1017	High	DeepSeek	28.6	157.2 t/s	$0.30 / $1.20
61	Step 5 Preview	—	StepFun	28.5	158.3 t/s	$1.00 / $2.70
62	GPT-6 Luna	Max	OpenAI	27.9	123.2 t/s	$0.10 / $0.50
63	GLM-5.3-Flash	—	Z AI	27.8	51.3 t/s	$0.15 / $0.50
64	MiMo-V2.6-Flash	—	Xiaomi	26.7	48.1 t/s	$0.14 / $0.28
65	Mistral Large 4	High	Mistral	26.1	137 t/s	$0.70 / $2.10
66	Claude Sonnet 5.5	Low	Anthropic	25.6	103.6 t/s	$2.00 / $10.00
67	Mistral Large 4 Preview	—	Mistral	25.3	131.6 t/s	$0.70 / $2.10
68	Muse Spark Open	Xhigh	Meta	24.9	222.9 t/s	$1.25 / $4.25
69	Claude Haiku 5.5	Medium	Anthropic	24.6	170.5 t/s	$0.10 / $0.50
70	GPT-5.6 Luna	Max	OpenAI	23.8	125.5 t/s	$0.20 / $1.20
71	GPT-6 Luna	Xhigh	OpenAI	23.5	117.7 t/s	$0.10 / $0.50
72	DeepSeek V4 Pro 0813	Max	DeepSeek	23.1	79.1 t/s	$1.32 / $3.96
72	GLM-5.3	Low	Z AI	23.1	77.3 t/s	$1.40 / $4.40
72	Mistral Large 4	Medium	Mistral	23.1	134.7 t/s	$0.70 / $2.10
75	Gemini 3.8 Flash	Low	Google	22.5	128.3 t/s	$0.75 / $3.75
76	GPT-6 Luna	High	OpenAI	21.4	120.2 t/s	$0.10 / $0.50
77	Nanbeige4.3-33B	—	Nanbeige	20.4	—	—
78	Qwen3.8 27B	Xhigh	Alibaba	20.1	38.2 t/s	$0.50 / $3.00
79	MiniMax-M3	—	MiniMax	18.9	136.6 t/s	$0.30 / $1.20
80	Claude Haiku 5.5	Low	Anthropic	18.6	165.9 t/s	$0.10 / $0.50
81	DeepSeek V4.1 Flash 1017	Non-reasoning	DeepSeek	18.5	148.2 t/s	$0.30 / $1.20
81	GPT-6 Luna	Medium	OpenAI	18.5	112.1 t/s	$0.10 / $0.50
83	K2 Horizon 375B A23B	—	Institute of Foundation Models	17.8	877 t/s	$0.00 / $0.00
84	Beam 1.0	—	Reflection	17.7	367.7 t/s	$0.40 / $1.20
85	Qwen3.8 27B	Medium	Alibaba	17.5	38.9 t/s	$0.50 / $3.00
86	Kimi K3	Low	Kimi	17.4	49.5 t/s	$3.00 / $15.00
87	Qwen3.8 27B	Low	Alibaba	15.7	39.3 t/s	$0.50 / $3.00
88	Nemotron 3.5 Super VL	—	NVIDIA	13.6	279.3 t/s	$0.17 / $0.50
89	Inkling Small	—	Thinking Machines	12.9	78.8 t/s	$0.30 / $1.20
90	DeepSeek V4 Pro 0813	Non-reasoning	DeepSeek	12.6	79.6 t/s	$1.32 / $3.96
91	Inkling	Xhigh	Thinking Machines	12.5	161.5 t/s	$1.00 / $4.05
92	Nemotron 3 Ultra 550B A55B	Reasoning	NVIDIA	11.8	253.8 t/s	$0.60 / $2.70
93	Gemini 3.5 Flash-Lite	—	Google	11.7	410 t/s	$0.30 / $2.50
94	Qwen3.8 27B	Non-reasoning	Alibaba	10	36 t/s	$0.50 / $3.00
95	GPT-6 Luna	Low	OpenAI	9.7	113.7 t/s	$0.10 / $0.50
96	GPT-6 Luna	Non-reasoning	OpenAI	9.4	111.1 t/s	$0.10 / $0.50
97	K-EXAONE 2.0	—	LG AI Research	7.6	—	—
98	Claude 4.5 Haiku	Reasoning	Anthropic	7.4	95.8 t/s	$1.00 / $5.00
99	Muse Glimmer	High	Meta	6.3	108.5 t/s	$0.30 / $1.20
100	Command A+	—	Cohere	6.2	258.1 t/s	$0.00 / $0.00
100	Gemini 3.1 Flash-Lite	—	Google	6.2	354.8 t/s	$0.25 / $1.50
102	Nemotron 3.5 Lightning	—	NVIDIA	5.5	286.9 t/s	$0.06 / $0.20
103	Granite 4.2 8B	—	IBM	5.1	139 t/s	$0.06 / $0.25
104	Granite 4.2 3B	—	IBM	4.7	—	—
105	Mistral Medium 3.5	—	Mistral	4.6	170.3 t/s	$1.50 / $7.50
106	Gemma 4 31B	Reasoning	Google	4.4	34.9 t/s	$0.00 / $0.00
107	Mistral Small 4	Reasoning	Mistral	4.3	161.2 t/s	$0.15 / $0.60
108	North Mini Code	—	Cohere	3.4	174.8 t/s	$0.00 / $0.00
109	NVIDIA Nemotron 3 Nano 4B	—	NVIDIA	1.8	—	—

数据点评

排名 1 · Gemini 4 Carbon · 指数 / 偏好分 60 ↗

榜内未收录本站条目的模型只显示名字，不做假链接。

02 / LMArena

人类偏好榜

最后更新 2026-11-09

数据说明

发布方: LMArena · 来源 ↗ · 本站不自创综合总分；两榜不合并、不换算。

照录 LMArena 官方公开数据集 lmarena-ai/leaderboard-dataset（CC-BY-4.0）文本总榜 overall 最新一期：共 421 个模型配置，官方发布日 2026-11-09。评分是 Bradley-Terry 模型算出的 Arena 评分（Elo 口径），名次、票数与置信区间均为官方原值，本站不改写、不加权。榜单含未收录进本站的模型，仅不可点击。完整官方数据：https://hf-mirror.com/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet

排名	模型	厂商	指数 / 偏好分	票数
1	Gemini 4 Carbon (High)	google	1538.3	4,282
2	Gemini 4 Argon (High)	google	1530.2	19,787
3	Claude Opus 5.5 (High)	anthropic	1513	15,249
4	Claude Fable 5.1 (Max)	anthropic	1512.4	20,087
5	Claude Opus 5 (Max)	anthropic	1506.3	35,710
6	Muse Wonder (Max)	meta	1503	12,032
7	Claude Opus 5 (High)	anthropic	1502.3	73,697
8	Claude Opus 4.6 (High)	anthropic	1501.7	84,424
9	Gemini 3.8 Flash (High)	google	1500.4	61,143
10	Claude Opus 4.6	anthropic	1496.7	89,228
11	MiMo-V2.6-Pro	xiaomi	1491.7	4,066
12	Claude Fable 5 (High)	anthropic	1491.6	45,524
13	Muse Spark 1.3 (Max)	meta	1491.5	26,111
14	Claude Opus 4.7 (High)	anthropic	1489.7	64,308
15	Gemini 3.7 Flash (High)	google	1488.8	33,300
16	Qwen3.8-Max	alibaba	1485.2	36,324
17	Muse Spark 1.2 (xHigh)	meta	1484	15,004
18	Claude Opus 4.7	anthropic	1483.2	65,369
19	Gemini 3.5 Flash (High)	google	1482.9	47,642
20	Gemini 3.1 Pro Preview	google	1480	132,296
21	Gemini 3 Pro	google	1479.5	41,829
22	Gemini 3.6 Flash (High)	google	1479.3	35,405
23	Muse Spark 1.1	meta	1478.7	45,562
24	Gemini 3.5 Flash (Medium)	google	1477.9	46,377
25	GLM-5.3 (Max)	zai	1475.9	31,017
26	Qwen3.7-Max-Preview	alibaba	1474.8	3,904
27	Kimi K3 (Max)	moonshot	1474.3	40,867
28	Muse Spark	meta	1474.1	14,009
29	GPT-5.5 (High)	openai	1471.2	68,359
30	Qwen3.5-Max-Preview	alibaba	1470.9	22,419

当前显示前 30 名，共 421 名官方数据集 ↗

数据点评

排名 1 · Gemini 4 Carbon (High) · 指数 / 偏好分 1538.3

03 / LMArena

更多品类榜

更多品类榜 · 21

LMArena 官方数据集共 22 个品类榜，下列每榜展示前 10 名。偏好榜为 Arena 评分（Elo 口径），Agent 系为 0–1 任务得分，量纲不同，只在同一榜内比较。

发布方: LMArena · CC-BY-4.0 · 官方数据集 ↗ · 最后更新 2026-11-10

文本·风格控制 · 421 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Gemini 4 Carbon (High)	1526
2	Gemini 4 Argon (High)	1522.4
3	Muse Wonder (Max)	1511
4	Claude Opus 5.5 (High)	1506.1
5	Claude Fable 5 (High)	1503.9
6	Claude Opus 4.6 (High)	1503
7	Claude Fable 5.1 (Max)	1502.7
8	Claude Opus 4.7 (High)	1501.1
9	Gemini 3.8 Flash (High)	1497.5
10	Claude Opus 4.6	1496.5

文生图 · 84 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	GPT-Image-2.5 Sunburst	1428.8
2	GPT-Image-2.6 Sunburst (Medium)	1406.1
3	GPT-Image-2.5 Flare	1400.2
4	GPT-Image-2.6 Flare (Medium)	1382.8
5	GPT-Image-2 (Medium)	1381.7
6	MAI-Image-2.6	1329.3
7	Gemini Nano Banana 2.1	1327.7
8	Grok Imagine Image 2.0	1318.6
9	Reve 2.1	1301.8
10	Grok Imagine Image 2.0 (20260801)	1288

视觉理解 · 167 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Claude Fable 5 (High)	1325.6
2	Claude Opus 5.5 (High)	1323.8
3	Claude Fable 5.1 (Max)	1321.6
4	Claude Opus 4.7	1318.7
5	Claude Opus 5 (High)	1317.5
6	Gemini 3.8 Flash (High)	1316.8
7	Claude Opus 4.6 (High)	1315.6
8	Claude Opus 4.7 (High)	1313.9
9	Gemini 3.7 Flash (High)	1313.1
10	Claude Opus 4.6	1312.2

文生视频 · 50 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Gemini Omni 1.2 Flash	1540.1
2	Gemini Omni 1.1 Flash	1513
3	gemini-omni-flash	1509.4
4	FLUX 3 Video	1491.1
5	Grok Imagine Video 1.5	1487.5
6	Dreamina Seedance 2.0 (720p)	1483.4
7	Dreamina Seedance 2.5 (720p)	1479.5
8	Wan 3.0	1476.2
9	SkyReels V4	1458.7
10	Muse Video	1457.1

图像编辑 · 63 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	GPT-Image-2.5 Sunburst	1519.8
2	GPT-Image-2.5 Flare	1484.5
3	GPT-Image-2.6 Sunburst (Medium)	1476.7
4	MAI-Image-2.7-Preview	1468.3
5	GPT-Image-2 (Medium)	1460.9
6	GPT-Image-2.6 Flare (Medium)	1460.1
7	Grok Imagine Image 2.0	1436.9
8	MAI-Image-2.6	1427.7
9	Gemini Nano Banana 2.1	1427.2
10	Grok Imagine Image 2.0 (20260801)	1418.1

视频编辑 · 11 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Wan 3.0	1424
2	Dreamina Seedance 2.5 (720p)	1417.2
3	Gemini Omni 1.2 Flash	1395.6
4	MiniMax H3	1383
5	gemini-omni-flash	1378
6	Dreamina Seedance 2.0 (720p)	1369.1
7	HappyHorse 1.0	1305.1
8	Kling O3 Pro	1262.5
9	Grok Imagine Video	1250.1
10	Kling O1 Pro	1189.6

Web 开发 · 154 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Claude Opus 5.5 (Max)	1817.7
2	Claude Sonnet 5.5 (xHigh)	1774.3
3	Gemini 4 Carbon (High)	1768
4	GPT-6 Astra (Max)	1766.8
5	Claude Fable 5.1 (Max)	1744.6
6	GPT-6.1 Sol (Max)	1742.8
7	Muse Wonder Full (Max)	1723.9
8	Claude Sonnet 5.5 (High)	1713.2
9	Claude Opus 5 (Max)	1690.8
10	Kimi K3.1 Pro (Max)	1687.8

搜索 · 42 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Qwen3.8-Max	1259.8
2	GPT-5.6 Sol (xHigh)	1257.8
3	Claude Opus 4.6 (Search)	1251.7
4	GPT-5.5 (Search)	1245.9
5	Claude Opus 5 (High)	1239
6	Claude Fable 5.1 (High)	1237.3
7	GPT-6 Astra (Max)	1230.8
8	GPT-6.1 Sol (Max)	1230.4
9	Claude Opus 4.7	1230.1
10	Claude Fable 5 (High)	1228.2

文档 · 46 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Claude Fable 5.1 (Max)	1514.8
2	Claude Opus 5 (High)	1508.7
3	Claude Opus 4.6 (High)	1507.6
4	Claude Opus 4.6	1506.2
5	Claude Fable 5 (High)	1496.6
6	Claude Opus 4.7 (High)	1495.7
7	Claude Opus 4.7	1494.8
8	GPT-5.5 (High)	1484.3
9	Gemini 3.8 Flash (High)	1481.9
10	Claude Sonnet 4.6	1481.9

Agent · 57 条目

任务得分 · 2026-11-08

#	模型	得分
1	Claude Opus 5.5 (Max)	0.200
2	Gemini 4 Carbon (High)	0.166
3	Claude Sonnet 5.5 (Max)	0.160
4	Claude Fable 5.1 (Max)	0.143
5	GPT-6 Astra (Max)	0.142
6	GPT-6.1 Sol (Max)	0.125
7	GPT-6.1 Terra (Max)	0.121
8	Gemini 4 Argon (High)	0.108
9	Muse Wonder Full (Max)	0.091
10	Claude Fable 5 (High)	0.077

文本·事实性 · 190 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Gemini 4 Carbon (High)	1521.2
2	Claude Opus 5.5 (High)	1505.5
3	Claude Fable 5.1 (Max)	1498.8
4	Gemini 4 Argon (High)	1497.8
5	Muse Wonder (Max)	1495.1
6	Claude Opus 4.6 (High)	1492.6
7	Claude Opus 4.6	1488
8	Claude Opus 5 (Max)	1487.6
9	GPT-5.5 (High)	1485.4
10	MiMo-V2.6-Pro	1484.9

搜索·风格控制 · 42 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	GPT-5.6 Sol (xHigh)	1255.4
2	GPT-6.1 Sol (Max)	1248.5
3	GPT-6 Astra (Max)	1244.8
4	GPT-6 Sol (Max)	1229
5	Claude Fable 5 (High)	1228.6
6	Qwen3.8-Max	1227.2
7	Claude Fable 5.1 (High)	1226.8
8	GPT-5.5 (Search)	1226
9	Claude Opus 4.6 (Search)	1221.7
10	Claude Opus 5 (High)	1213.3

搜索·事实性 · 40 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	GPT-5.6 Sol (xHigh)	1249.5
2	GPT-5.5 (Search)	1246.3
3	Qwen3.8-Max	1242.9
4	GPT-6 Astra (Max)	1238.7
5	GPT-6.1 Sol (Max)	1236.5
6	Claude Opus 4.6 (Search)	1227.4
7	GPT-5.2 (Search)	1224
8	Grok 4.5	1220
9	Claude Fable 5.1 (High)	1214.4
10	ERNIE 5.1	1213.8

文档·风格控制 · 46 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Claude Fable 5 (High)	1500.1
2	Claude Fable 5.1 (Max)	1498.4
3	Claude Opus 4.6 (High)	1496.9
4	Claude Opus 4.6	1495.6
5	Claude Opus 4.7	1494.4
6	Claude Opus 4.7 (High)	1494.3
7	GPT-5.6 Sol (xHigh)	1486.5
8	Claude Opus 5 (High)	1485.9
9	Claude Opus 4.8 (High)	1485.4
10	GPT-5.5 (High)	1481.4

视觉·风格控制 · 167 条目

指数 / 偏好分 · 2026-11-09

#	模型	指数 / 偏好分
1	Claude Fable 5 (High)	1309.4
2	Claude Opus 4.7	1300.3
3	Qwen3.8-Max	1299
4	Claude Opus 4.6 (High)	1298.8
5	Claude Opus 4.7 (High)	1298.4
6	Claude Opus 5.5 (High)	1295.3
7	Muse Spark	1294.4
8	Gemini 3.7 Flash (High)	1293.8
9	Muse Spark 1.2 (xHigh)	1293.5
10	Claude Opus 4.6	1293.2

Agent·错误恢复 · 52 条目

任务得分 · 2026-10-08

#	模型	得分
1	Claude Sonnet 5.5 (Max)	0.150
2	Claude Opus 5 (Max)	0.136
3	Claude Opus 5.5 (High)	0.136
4	Claude Opus 5 (High)	0.126
5	Claude Fable 5.1 (Max)	0.116
6	Claude Fable 5 (High)	0.096
7	GPT 5.5 (xHigh)	0.095
8	Deepseek V4.1 Flash (Max)	0.079
9	Claude Sonnet 5 (High)	0.074
10	Claude Opus 4.8 (High)	0.072

Agent·可靠性 · 57 条目

任务得分 · 2026-11-08

#	模型	得分
1	Claude Sonnet 5 (High)	0.150
2	Claude Opus 5.5 (Max)	0.150
3	GPT-6.1 Terra (Max)	0.133
4	Claude Sonnet 5.5 (Max)	0.127
5	GPT-5.6 Sol (xHigh)	0.113
6	GPT-6 Sol (Max)	0.110
7	Claude Fable 5 (High)	0.104
8	Grok 4.6 (xHigh)	0.102
9	Claude Fable 5.1 (Max)	0.100
10	Gemini 3.8 Flash (High)	0.099

Agent·反馈 · 57 条目

任务得分 · 2026-11-08

#	模型	得分
1	Claude Opus 5.5 (Max)	0.423
2	Gemini 4 Carbon (High)	0.395
3	Claude Sonnet 5.5 (Max)	0.316
4	GPT-6 Astra (Max)	0.309
5	Claude Fable 5.1 (Max)	0.298
6	GPT-6.1 Sol (Max)	0.284
7	Gemini 4 Argon (High)	0.235
8	GPT-6.1 Terra (Max)	0.204
9	Muse Wonder Full (Max)	0.154
10	Claude Fable 5 (High)	0.146

Agent·可控性 · 57 条目

任务得分 · 2026-11-08

#	模型	得分
1	Gemini 4 Carbon (High)	0.191
2	Claude Opus 5.5 (Max)	0.186
3	GPT-6.1 Terra (Max)	0.158
4	Claude Fable 5.1 (Max)	0.157
5	GPT-6 Astra (Max)	0.148
6	Claude Sonnet 5.5 (Max)	0.138
7	Gemini 4 Argon (High)	0.135
8	GPT-6.1 Sol (Max)	0.133
9	Muse Wonder Full (Max)	0.104
10	GPT-6 Sol (Max)	0.101

Agent·工具幻觉 · 57 条目

任务得分 · 2026-11-08

#	模型	得分
1	GLM-5.3-Flash	0.009
2	GPT-6 Astra (Max)	0.009
3	GPT-6.1 Sol (Max)	0.009
4	Kimi K3.1 Pro (Max)	0.009
5	GPT-6.1 Terra (Max)	0.009
6	GPT-6 Luna (Low)	0.009
7	GPT-6 Luna (Medium)	0.009
8	Grok 4.6 (xHigh)	0.009
9	GPT-5.6 Sol (xHigh)	0.009
10	GLM-5.3 (Max)	0.009

Agent·任务结果 · 57 条目

任务得分 · 2026-11-08

#	模型	得分
1	Claude Opus 5.5 (Max)	0.207
2	Claude Sonnet 5.5 (Max)	0.195
3	Gemini 4 Carbon (High)	0.184
4	Gemini 4 Argon (High)	0.141
5	GPT-6 Astra (Max)	0.136
6	Claude Fable 5.1 (Max)	0.131
7	GPT-6.1 Sol (Max)	0.112
8	Muse Wonder Full (Max)	0.110
9	GPT-6.1 Terra (Max)	0.106
10	Claude Opus 5 (Max)	0.105