1
A
claude-opus-4-6-thinking
claude-opus-4-6-thinkingAnthropic
100.0
11.9K
1M
¥36 / ¥180输入/输出
对话 · 文本 · 长程任务排行榜
基于公开偏好评测数据展示 文本 / 长程任务 榜单，每个榜单独立排序。
长程任务模型怎么选？
长程任务排行榜用于回答“长程任务哪个 AI 模型更强”这类选型问题。页面基于公开偏好评测数据展示模型排名，每个榜单独立排序，不把不同任务的原始分数直接混合。当前榜首是 claude-opus-4-6-thinking，厂商为 anthropic，百分制分数 100.0，样本量 11.9K。 数据来源保留 Arena 原始榜单和 LMArena leaderboard dataset，更新时间以榜单发布时间 2026-05-27 为准。
claude-opus-4-6-thinkingclaude-opus-4-6claude-opus-4-7-thinkingclaude-opus-4-7gpt-5.5-high
2
Aclaude-opus-4-6
claude-opus-4-6Anthropic
99.7
13.1K
1M
¥36 / ¥180输入/输出
3
Aclaude-opus-4-7-thinking
claude-opus-4-7-thinkingAnthropic
99.4
8.3K
1M
¥36 / ¥180输入/输出
4
Aclaude-opus-4-7
claude-opus-4-7Anthropic
99.1
8.7K
1M
¥36 / ¥180输入/输出
5
O
gpt-5.5-high
gpt-5.5-highOpenai
98.8
6.8K
1.05M
¥36 / ¥216输入/输出
6
A
qwen3.7-max-preview
qwen3.7-max-previewAlibaba
98.5
1.6K
1M
¥18 / ¥54输入/输出
7
G
gemini-3.1-pro-preview
gemini-3.1-pro-previewGoogle
98.2
15.9K
1.05M
¥14.4 / ¥86.4输入/输出
8
Aclaude-sonnet-4-6
claude-sonnet-4-6Anthropic
97.9
10K
1M
¥21.6 / ¥108输入/输出
9
MI
mimo-v2.5-pro
mimo-v2.5-proXiaomi
97.6
6.3K
1.05M
¥7.2 / ¥21.6输入/输出
10
Ggemini-3.5-flash
gemini-3.5-flashGoogle
97.3
3.7K
1.05M
¥10.8 / ¥64.8输入/输出
11
Ogpt-5.4-high
gpt-5.4-highOpenai
97.0
10.6K
1.05M
¥18 / ¥108输入/输出
12
Aclaude-opus-4-5-20251101
claude-opus-4-5-20251101Anthropic
96.7
20.1K
200K
¥36 / ¥180输入/输出
13
Aclaude-opus-4-5-20251101-thinking-32k
claude-opus-4-5-20251101-thinking-32kAnthropic
96.4
9.3K
200K
¥108 / ¥540输入/输出
14
Aclaude-sonnet-4-5-20250929
claude-sonnet-4-5-20250929Anthropic
96.1
22.4K
200K
¥21.6 / ¥108输入/输出
15
Z
glm-5.1
glm-5.1Zai
95.8
5.4K
200K
¥0 / ¥0输入/输出
16
Aqwen3.5-max-preview
qwen3.5-max-previewAlibaba
95.5
7.4K
-
-
17
Ogpt-5.5
gpt-5.5Openai
95.3
7K
1.05M
¥36 / ¥216输入/输出
18
Aclaude-sonnet-4-5-20250929-thinking-32k
claude-sonnet-4-5-20250929-thinking-32kAnthropic
95.0
22.1K
200K
¥21.6 / ¥108输入/输出
19
Ggemini-3-pro
gemini-3-proGoogle
94.7
10.5K
1.05M
¥14.4 / ¥86.4输入/输出
20
Ogpt-5.4
gpt-5.4Openai
94.4
11.3K
1.05M
¥18 / ¥108输入/输出
21
B
ernie-5.1
ernie-5.1Baidu
94.1
5.5K
119K
¥5.4 / ¥21.6输入/输出
22
M
kimi-k2.6
kimi-k2.6Moonshot
93.8
6.1K
262K
¥6.84 / ¥28.8输入/输出
23
Aclaude-opus-4-1-20250805-thinking-16k
claude-opus-4-1-20250805-thinking-16kAnthropic
93.5
11.1K
200K
¥108 / ¥540输入/输出
24
MImimo-v2-pro
mimo-v2-proXiaomi
93.2
8.2K
1.05M
¥7.2 / ¥21.6输入/输出
25
Aqwen3.6-max-preview
qwen3.6-max-previewAlibaba
92.9
1.8K
246K
¥9.5 / ¥56.9输入/输出
26
D
deepseek-v4-pro-thinking
deepseek-v4-pro-thinkingDeepseek
92.6
6.4K
1M
¥3.13 / ¥6.26输入/输出
27
Ddeepseek-v4-pro
deepseek-v4-proDeepseek
92.3
6.8K
1M
¥3.13 / ¥6.26输入/输出
28
Ggemini-3-flash
gemini-3-flashGoogle
92.0
7.9K
1.05M
¥3.6 / ¥21.6输入/输出
29
Ggemini-2.5-pro
gemini-2.5-proGoogle
91.7
31.2K
1.05M
¥9 / ¥72输入/输出
30
Zglm-5
glm-5Zai
91.4
7.2K
205K
¥7.2 / ¥23输入/输出
31
M
muse-spark
muse-sparkMeta
91.1
4.4K
-
-
32
Ogpt-5.1-high
gpt-5.1-highOpenai
90.8
10.3K
400K
¥9 / ¥72输入/输出
33
MImimo-v2.5
mimo-v2.5Xiaomi
90.5
6.7K
1.05M
¥2.88 / ¥14.4输入/输出
34
Aclaude-opus-4-1-20250805
claude-opus-4-1-20250805Anthropic
90.2
17.7K
200K
¥108 / ¥540输入/输出
35
Ggemma-4-31b
gemma-4-31bGoogle
89.9
1.6K
262K
¥3.24 / ¥7.2输入/输出
36
Aqwen3.5-397b-a17b
qwen3.5-397b-a17bAlibaba
89.6
11.6K
262K
¥3.1 / ¥18.6输入/输出
37
Mkimi-k2.5-thinking
kimi-k2.5-thinkingMoonshot
89.3
12K
262K
¥4.32 / ¥21.6输入/输出
38
X
grok-3-preview-02-24
grok-3-preview-02-24Xai
89.0
4.9K
1M
¥9 / ¥18输入/输出
39
Aqwen3-max-preview
qwen3-max-previewAlibaba
88.7
6K
262K
¥6.2 / ¥24.8输入/输出
40
Aqwen3.6-plus
qwen3.6-plusAlibaba
88.4
7.2K
1M
¥3.6 / ¥21.6输入/输出
41
Mkimi-k2.5-instant
kimi-k2.5-instantMoonshot
88.1
2.2K
262K
¥4.32 / ¥21.6输入/输出
42
A
amazon-nova-experimental-chat-26-02-10
amazon-nova-experimental-chat-26-02-10Amazon
87.8
947
-
-
43
Zglm-4.7
glm-4.7Zai
87.5
3.1K
205K
¥0 / ¥0输入/输出
44
Xgrok-4.20-beta-0309-reasoning
grok-4.20-beta-0309-reasoningXai
87.2
10.9K
2M
¥14.4 / ¥43.2输入/输出
45
Ggemini-3-flash (thinking-minimal)
gemini-3-flash (thinking-minimal)Google
86.9
17.4K
1.05M
¥3.6 / ¥21.6输入/输出
46
Ogpt-5.1
gpt-5.1Openai
86.6
11.5K
400K
¥9 / ¥72输入/输出
47
Xgrok-4.20-multi-agent-beta-0309
grok-4.20-multi-agent-beta-0309Xai
86.4
10.7K
2M
¥14.4 / ¥43.2输入/输出
48
Ddeepseek-v4-flash
deepseek-v4-flashDeepseek
86.1
6.9K
1M
¥1.01 / ¥2.02输入/输出
49
Ggemma-4-26b-a4b
gemma-4-26b-a4bGoogle
85.8
1.5K
262K
¥0.94 / ¥2.88输入/输出
50
B
dola-seed-2.0-pro
dola-seed-2.0-proBytedance
85.5
13.1K
-
-
51
Aclaude-haiku-4-5-20251001
claude-haiku-4-5-20251001Anthropic
85.2
22.9K
200K
¥7.2 / ¥36输入/输出
52
Ogpt-5.2-chat-latest-20260210
gpt-5.2-chat-latest-20260210Openai
84.9
11.5K
400K
¥12.6 / ¥101输入/输出
53
Ddeepseek-v3.2
deepseek-v3.2Deepseek
84.6
12.2K
128K
¥2.09 / ¥3.1输入/输出
54
Aqwen3-235b-a22b-instruct-2507
qwen3-235b-a22b-instruct-2507Alibaba
84.3
24.6K
128K
¥2.09 / ¥8.23输入/输出
55
Xgrok-4.20-beta1
grok-4.20-beta1Xai
84.0
8.2K
2M
¥14.4 / ¥43.2输入/输出
56
Ddeepseek-v3.2-thinking
deepseek-v3.2-thinkingDeepseek
83.7
10.4K
128K
¥2.09 / ¥3.1输入/输出
57
Zglm-4.6
glm-4.6Zai
83.4
8.8K
205K
¥4.32 / ¥15.8输入/输出
58
Ddeepseek-v4-flash-thinking
deepseek-v4-flash-thinkingDeepseek
83.1
6.8K
1M
¥1.01 / ¥2.02输入/输出
59
Bernie-5.0-0110
ernie-5.0-0110Baidu
82.8
10.3K
128K
¥7.92 / ¥14.4输入/输出
60
M
longcat-flash-chat-2602-exp
longcat-flash-chat-2602-expMeituan
82.5
8.5K
128K
¥1.08 / ¥10.8输入/输出
61
Aqwen3-vl-235b-a22b-instruct
qwen3-vl-235b-a22b-instructAlibaba
82.2
2.2K
128K
¥2.16 / ¥8.64输入/输出
62
Ddeepseek-v3.1-thinking
deepseek-v3.1-thinkingDeepseek
81.9
2.4K
128K
¥1.44 / ¥5.04输入/输出
63
Ogpt-5.5-instant
gpt-5.5-instantOpenai
81.6
11.1K
400K
¥9 / ¥72输入/输出
64
Aclaude-opus-4-20250514-thinking-16k
claude-opus-4-20250514-thinking-16kAnthropic
81.3
6.9K
200K
¥108 / ¥540输入/输出
65
Ddeepseek-v3.1-terminus-thinking
deepseek-v3.1-terminus-thinkingDeepseek
81.0
680
128K
¥1.8 / ¥5.04输入/输出
66
MImimo-v2-omni
mimo-v2-omniXiaomi
80.7
1.3K
262K
¥2.88 / ¥14.4输入/输出
67
Ggemini-2.5-flash
gemini-2.5-flashGoogle
80.4
30.5K
1.05M
¥2.16 / ¥18输入/输出
68
Ddeepseek-v3.2-exp
deepseek-v3.2-expDeepseek
80.1
2.9K
128K
¥0 / ¥0输入/输出
69
Xgrok-4-fast-chat
grok-4-fast-chatXai
79.8
1.5K
2M
¥1.44 / ¥3.6输入/输出
70
M
minimax-m2.1-preview
minimax-m2.1-previewMinimax
79.5
4.3K
205K
¥0 / ¥0输入/输出
71
Mminimax-m2.7
minimax-m2.7Minimax
79.2
8.4K
205K
¥0 / ¥0输入/输出
72
Xgrok-4.1
grok-4.1Xai
78.9
18.6K
200K
¥14.4 / ¥72输入/输出
73
Mkimi-k2-thinking-turbo
kimi-k2-thinking-turboMoonshot
78.6
17.5K
262K
¥17.3 / ¥72输入/输出
74
Ochatgpt-4o-latest-20250326
chatgpt-4o-latest-20250326Openai
78.3
17.3K
128K
¥18 / ¥72输入/输出
75
Ogpt-5.2
gpt-5.2Openai
78.0
15.7K
400K
¥12.6 / ¥101输入/输出
76
Aqwen3-max-2025-09-23
qwen3-max-2025-09-23Alibaba
77.7
2.1K
258K
¥6.19 / ¥24.7输入/输出
77
Aqwen3.5-27b
qwen3.5-27bAlibaba
77.4
9.2K
262K
¥2.16 / ¥17.3输入/输出
78
MA
mistral-large-3
mistral-large-3Mistral
77.2
11.9K
262K
¥3.6 / ¥10.8输入/输出
79
Zglm-4.5
glm-4.5Zai
76.9
5K
131K
¥4.32 / ¥15.8输入/输出
80
TE
hunyuan-hy3-preview
hunyuan-hy3-previewTencent
76.6
2.3K
256K
¥0 / ¥0输入/输出
81
Xgrok-4-fast-reasoning
grok-4-fast-reasoningXai
76.3
4.5K
2M
¥1.44 / ¥3.6输入/输出
82
Ogpt-5.2-high
gpt-5.2-highOpenai
76.0
14.9K
400K
¥12.6 / ¥101输入/输出
83
Xgrok-4-0709
grok-4-0709Xai
75.7
8.8K
256K
¥21.6 / ¥108输入/输出
84
Xgrok-4.1-thinking
grok-4.1-thinkingXai
75.4
18.6K
200K
¥14.4 / ¥72输入/输出
85
Bernie-5.0-preview-1203
ernie-5.0-preview-1203Baidu
75.1
2.5K
128K
¥7.92 / ¥14.4输入/输出
86
Ogpt-5.4-mini-high
gpt-5.4-mini-highOpenai
74.8
10.2K
400K
¥5.4 / ¥32.4输入/输出
87
MImimo-v2-flash (non-thinking)
mimo-v2-flash (non-thinking)Xiaomi
74.5
14K
262K
¥0.72 / ¥2.16输入/输出
88
MAmistral-medium-2508
mistral-medium-2508Mistral
74.2
25.2K
262K
¥2.88 / ¥14.4输入/输出
89
Ddeepseek-v3.2-exp-thinking
deepseek-v3.2-exp-thinkingDeepseek
73.9
2.1K
128K
¥0 / ¥0输入/输出
90
Bernie-5.0-preview-1022
ernie-5.0-preview-1022Baidu
73.6
1.2K
128K
¥7.92 / ¥14.4输入/输出
91
Aqwen3.5-122b-a10b
qwen3.5-122b-a10bAlibaba
73.3
9.6K
262K
¥2.88 / ¥23输入/输出
92
Ggemini-2.5-flash-preview-09-2025
gemini-2.5-flash-preview-09-2025Google
73.0
7.9K
1M
¥2.16 / ¥18输入/输出
93
Ogpt-4.5-preview-2025-02-27
gpt-4.5-preview-2025-02-27Openai
72.7
1.8K
8.19K
¥216 / ¥432输入/输出
94
ST
step-3.5-flash
step-3.5-flashStepfun
72.4
11.3K
256K
¥0.69 / ¥2.07输入/输出
95
Aqwen3-next-80b-a3b-instruct
qwen3-next-80b-a3b-instructAlibaba
72.1
5.2K
131K
¥1.04 / ¥4.13输入/输出
96
Xgrok-4.3
grok-4.3Xai
71.8
6.6K
1M
¥9 / ¥18输入/输出
97
Aclaude-opus-4-20250514
claude-opus-4-20250514Anthropic
71.5
8.4K
200K
¥108 / ¥540输入/输出
98
Aamazon-nova-experimental-chat-26-01-10
amazon-nova-experimental-chat-26-01-10Amazon
71.2
886
-
-
99
Ogpt-5-chat
gpt-5-chatOpenai
70.9
6.7K
400K
¥9 / ¥72输入/输出
100
Ddeepseek-v3.1
deepseek-v3.1Deepseek
70.6
3.2K
128K
¥1.44 / ¥5.04输入/输出
101
TEhunyuan-vision-1.5-thinking
hunyuan-vision-1.5-thinkingTencent
70.3
506
-
-
102
Aqwen3-235b-a22b-thinking-2507
qwen3-235b-a22b-thinking-2507Alibaba
70.0
1.6K
131K
¥2.07 / ¥8.26输入/输出
103
Aclaude-sonnet-4-20250514-thinking-32k
claude-sonnet-4-20250514-thinking-32kAnthropic
69.7
6.7K
200K
¥21.6 / ¥108输入/输出
104
Aamazon-nova-experimental-chat-11-10
amazon-nova-experimental-chat-11-10Amazon
69.4
6.1K
-
-
105
Ddeepseek-v3.1-terminus
deepseek-v3.1-terminusDeepseek
69.1
846
128K
¥1.8 / ¥5.04输入/输出
106
MImimo-v2-flash (thinking)
mimo-v2-flash (thinking)Xiaomi
68.8
2.8K
262K
¥0.72 / ¥2.16输入/输出
107
Ggemini-3.1-flash-lite-preview
gemini-3.1-flash-lite-previewGoogle
68.5
12.7K
1.05M
¥1.8 / ¥10.8输入/输出
108
Ogpt-5.3-chat-latest
gpt-5.3-chat-latestOpenai
68.2
11K
128K
¥12.6 / ¥101输入/输出
109
Aqwen3-vl-235b-a22b-thinking
qwen3-vl-235b-a22b-thinkingAlibaba
68.0
1.7K
131K
¥2.06 / ¥8.26输入/输出
110
Aqwen3.5-35b-a3b
qwen3.5-35b-a3bAlibaba
67.7
9.9K
262K
¥1.8 / ¥14.4输入/输出
111
Aqwen3.5-flash
qwen3.5-flashAlibaba
67.4
10.5K
1M
¥1.24 / ¥12.4输入/输出
112
Ddeepseek-r1-0528
deepseek-r1-0528Deepseek
67.1
3.1K
164K
¥3.6 / ¥15.5输入/输出
113
Xgrok-4-1-fast-reasoning
grok-4-1-fast-reasoningXai
66.8
16K
2M
¥1.44 / ¥3.6输入/输出
114
Mlongcat-flash-chat
longcat-flash-chatMeituan
66.5
2.3K
128K
¥1.08 / ¥10.8输入/输出
115
Aqwen3-235b-a22b-no-thinking
qwen3-235b-a22b-no-thinkingAlibaba
66.2
7.1K
131K
¥2.07 / ¥8.26输入/输出
116
Aamazon-nova-experimental-chat-12-10
amazon-nova-experimental-chat-12-10Amazon
65.9
760
-
-
117
Ogpt-5-high
gpt-5-highOpenai
65.6
6.8K
400K
¥9 / ¥72输入/输出
118
TEhunyuan-t1-20250711
hunyuan-t1-20250711Tencent
65.3
889
131K
¥0 / ¥0输入/输出
119
Ogpt-4.1-2025-04-14
gpt-4.1-2025-04-14Openai
65.0
9.8K
1.05M
¥14.4 / ¥57.6输入/输出
120
Aclaude-sonnet-4-20250514
claude-sonnet-4-20250514Anthropic
64.7
7.7K
200K
¥21.6 / ¥108输入/输出
121
Aqwen3-coder-480b-a35b-instruct
qwen3-coder-480b-a35b-instructAlibaba
64.4
4.9K
262K
¥6.2 / ¥24.8输入/输出
122
Aqwen3-30b-a3b-instruct-2507
qwen3-30b-a3b-instruct-2507Alibaba
64.1
4.9K
262K
¥2.16 / ¥3.6输入/输出
123
Oo1-2024-12-17
o1-2024-12-17Openai
63.8
3.6K
128K
¥108 / ¥432输入/输出
124
Ggemini-2.5-flash-lite-preview-06-17-thinking
gemini-2.5-flash-lite-preview-06-17-thinkingGoogle
63.5
6.3K
65.5K
¥0.72 / ¥2.88输入/输出
125
Aclaude-3-7-sonnet-20250219-thinking-32k
claude-3-7-sonnet-20250219-thinking-32kAnthropic
63.2
6.2K
-
-
126
Xgrok-3-mini-high
grok-3-mini-highXai
62.9
3.3K
128K
¥0 / ¥0输入/输出
127
Ggemini-2.5-flash-lite-preview-09-2025-no-thinking
gemini-2.5-flash-lite-preview-09-2025-no-thinkingGoogle
62.6
11.5K
1.05M
¥0.72 / ¥2.88输入/输出
128
Oo3-2025-04-16
o3-2025-04-16Openai
62.3
11.5K
200K
¥14.4 / ¥57.6输入/输出
129
Aamazon-nova-experimental-chat-10-20
amazon-nova-experimental-chat-10-20Amazon
62.0
2.6K
-
-
130
NV
nvidia-nemotron-3-super-120b-a12b
nvidia-nemotron-3-super-120b-a12bNvidia
61.7
2K
262K
¥1.44 / ¥5.76输入/输出
131
Ogpt-5.4-nano-high
gpt-5.4-nano-highOpenai
61.4
10K
400K
¥1.44 / ¥9输入/输出
132
Zglm-4.5-air
glm-4.5-airZai
61.1
6.6K
131K
¥0 / ¥0输入/输出
133
Mminimax-m2.5
minimax-m2.5Minimax
60.8
12.6K
205K
¥0 / ¥0输入/输出
134
TEhunyuan-turbos-20250416
hunyuan-turbos-20250416Tencent
60.5
1.5K
131K
¥0 / ¥0输入/输出
135
MAmistral-medium-2505
mistral-medium-2505Mistral
60.2
5.9K
262K
¥2.88 / ¥14.4输入/输出
136
Zglm-4.6v
glm-4.6vZai
59.9
663
128K
¥2.16 / ¥6.48输入/输出
137
Aqwen2.5-max
qwen2.5-maxAlibaba
59.6
4.5K
32K
¥11.5 / ¥46输入/输出
138
Xgrok-3-mini-beta
grok-3-mini-betaXai
59.3
4.1K
1M
¥9 / ¥18输入/输出
139
Ogpt-5-mini-high
gpt-5-mini-highOpenai
59.1
5.8K
400K
¥1.8 / ¥14.4输入/输出
140
Ddeepseek-r1
deepseek-r1Deepseek
58.8
2.3K
164K
¥5.04 / ¥18输入/输出
141
Aclaude-3-7-sonnet-20250219
claude-3-7-sonnet-20250219Anthropic
58.5
6.7K
200K
¥21.6 / ¥108输入/输出
142
UNtrinity-large-preview
-
58.2
9.7K
262K
¥1.8 / ¥6.48输入/输出
143
Ddeepseek-v3-0324
deepseek-v3-0324Deepseek
57.9
8.5K
75K
¥1.44 / ¥5.76输入/输出
144
Aqwen3-next-80b-a3b-thinking
qwen3-next-80b-a3b-thinkingAlibaba
57.6
2.8K
131K
¥1.04 / ¥10.3输入/输出
145
Aqwen3-235b-a22b
qwen3-235b-a22bAlibaba
57.3
4.6K
131K
¥2.07 / ¥8.26输入/输出
146
Mkimi-k2-0905-preview
kimi-k2-0905-previewMoonshot
57.0
2.5K
262K
¥4.32 / ¥18输入/输出
147
Zglm-4.7-flash
glm-4.7-flashZai
56.7
3.2K
200K
¥0 / ¥0输入/输出
148
STstep-1o-turbo-202506
step-1o-turbo-202506Stepfun
56.4
1.3K
-
-
149
Ggemini-2.0-flash-001
gemini-2.0-flash-001Google
56.1
6.5K
1.05M
¥1.08 / ¥4.32输入/输出
150
Oo1-preview
o1-previewOpenai
55.8
4.6K
128K
¥108 / ¥432输入/输出
151
Ogpt-4.1-mini-2025-04-14
gpt-4.1-mini-2025-04-14Openai
55.5
7K
1.05M
¥2.88 / ¥11.5输入/输出
152
Ddeepseek-v3
deepseek-v3Deepseek
55.2
3.1K
128K
¥0 / ¥0输入/输出
153
Oo3-mini-high
o3-mini-highOpenai
54.9
2.2K
200K
¥7.92 / ¥31.7输入/输出
154
TEhunyuan-large-2025-02-10
hunyuan-large-2025-02-10Tencent
54.6
371
-
-
155
Anova-2-lite
nova-2-liteAmazon
54.3
2.9K
128K
¥2.38 / ¥19.8输入/输出
156
Mminimax-m2
minimax-m2Minimax
54.0
1.6K
197K
¥0 / ¥0输入/输出
157
CO
command-a-03-2025
command-a-03-2025Cohere
53.7
10.5K
256K
¥18 / ¥72输入/输出
158
Ggemma-3-27b-it
gemma-3-27b-itGoogle
53.4
6.6K
128K
¥2.15 / ¥2.15输入/输出
159
Aamazon-nova-experimental-chat-10-09
amazon-nova-experimental-chat-10-09Amazon
53.1
575
-
-
160
MAmistral-small-2506
mistral-small-2506Mistral
52.8
3.4K
262K
¥2.88 / ¥14.4输入/输出
161
Mminimax-m1
minimax-m1Minimax
52.5
7K
1M
¥0.95 / ¥9.03输入/输出
162
UNintellect-3
-
52.2
1.3K
131K
¥1.44 / ¥7.92输入/输出
163
IA
mercury-2
mercury-2Inception Ai
51.9
844
128K
¥1.8 / ¥5.4输入/输出
164
Aqwen3-32b
qwen3-32bAlibaba
51.6
504
131K
¥2.07 / ¥8.26输入/输出
165
Oo3-mini
o3-miniOpenai
51.3
9.2K
200K
¥7.92 / ¥31.7输入/输出
166
STstep-3
step-3Stepfun
51.0
1.3K
65.5K
¥1.8 / ¥4.68输入/输出
167
Mkimi-k2-0711-preview
kimi-k2-0711-previewMoonshot
50.7
5.5K
131K
¥4.32 / ¥18输入/输出
168
Aqwen-plus-0125
qwen-plus-0125Alibaba
50.4
690
1M
¥0.83 / ¥2.07输入/输出
169
AG
ling-flash-2.0
ling-flash-2.0Ant Group
50.1
1.3K
131K
¥1.01 / ¥4.1输入/输出
170
UNtrinity-large-thinking
-
49.9
9.5K
262K
¥1.8 / ¥6.48输入/输出
171
Ggemini-2.0-flash-lite-preview-02-05
gemini-2.0-flash-lite-preview-02-05Google
49.6
2.9K
1.05M
¥0.54 / ¥2.16输入/输出
172
Oo1-mini
o1-miniOpenai
49.3
7.9K
128K
¥7.92 / ¥31.7输入/输出
173
TEhunyuan-turbos-20250226
hunyuan-turbos-20250226Tencent
49.0
287
131K
¥0 / ¥0输入/输出
174
AGring-flash-2.0
ring-flash-2.0Ant Group
48.7
1.3K
131K
¥1.01 / ¥4.1输入/输出
175
Ogpt-oss-120b
gpt-oss-120bOpenai
48.4
6.5K
131K
¥1.08 / ¥4.32输入/输出
176
Ggemma-3-12b-it
gemma-3-12b-itGoogle
48.1
371
128K
¥1.96 / ¥1.96输入/输出
177
Zglm-4-plus-0111
glm-4-plus-0111Zai
47.8
765
128K
¥72 / ¥72输入/输出
178
NVnvidia-llama-3.3-nemotron-super-49b-v1.5
nvidia-llama-3.3-nemotron-super-49b-v1.5Nvidia
47.5
642
131K
¥2.88 / ¥2.88输入/输出
179
Oo4-mini-2025-04-16
o4-mini-2025-04-16Openai
47.2
8.5K
200K
¥7.92 / ¥31.7输入/输出
180
Aqwen3-30b-a3b
qwen3-30b-a3bAlibaba
46.9
4.5K
128K
¥0.79 / ¥7.78输入/输出
181
AI
olmo-3.1-32b-instruct
olmo-3.1-32b-instructAllenai
46.6
2.9K
200K
¥14.4 / ¥57.6输入/输出
182
Aclaude-3-5-sonnet-20241022
claude-3-5-sonnet-20241022Anthropic
46.3
13.8K
200K
¥21.6 / ¥108输入/输出
183
Ogpt-5-nano-high
gpt-5-nano-highOpenai
46.0
1.6K
400K
¥0.36 / ¥2.88输入/输出
184
Aqwq-32b
qwq-32bAlibaba
45.7
4K
131K
¥2.07 / ¥6.2输入/输出
185
TEhunyuan-turbo-0110
hunyuan-turbo-0110Tencent
45.4
313
-
-
186
Ggemini-1.5-pro-002
gemini-1.5-pro-002Google
45.1
8.4K
-
-
187
STstep-2-16k-exp-202412
step-2-16k-exp-202412Stepfun
44.8
693
16.4K
¥37.5 / ¥118输入/输出
188
Zglm-4.5v
glm-4.5vZai
44.5
1.1K
64K
¥4.32 / ¥13输入/输出
189
NVnvidia-nemotron-3-nano-30b-a3b-bf16
nvidia-nemotron-3-nano-30b-a3b-bf16Nvidia
44.2
3.9K
131K
¥0 / ¥0输入/输出
190
Ddeepseek-v2.5-1210
deepseek-v2.5-1210Deepseek
43.9
1K
1M
¥1.01 / ¥2.02输入/输出
191
TEhunyuan-standard-2025-02-10
hunyuan-standard-2025-02-10Tencent
43.6
432
-
-
192
NVllama-3.3-nemotron-49b-super-v1
llama-3.3-nemotron-49b-super-v1Nvidia
43.3
317
131K
¥0 / ¥0输入/输出
193
MAmistral-small-3.1-24b-instruct-2503
mistral-small-3.1-24b-instruct-2503Mistral
43.0
6.3K
262K
¥2.88 / ¥14.4输入/输出
194
NVllama-3.1-nemotron-ultra-253b-v1
llama-3.1-nemotron-ultra-253b-v1Nvidia
42.7
321
128K
¥4.32 / ¥13输入/输出
195
UNyi-lightning
-
42.4
3.6K
12K
¥1.44 / ¥1.44输入/输出
196
MAmagistral-medium-2506
magistral-medium-2506Mistral
42.1
2.4K
128K
¥14.4 / ¥36输入/输出
197
AIolmo-3-32b-think
olmo-3-32b-thinkAllenai
41.8
1.3K
128K
¥2.16 / ¥3.24输入/输出
198
Aqwen2.5-plus-1127
qwen2.5-plus-1127Alibaba
41.5
1.6K
-
-
199
Ggemini-1.5-pro-001
gemini-1.5-pro-001Google
41.2
9.8K
-
-
200
UNathene-v2-chat
-
40.9
3.7K
-
-
201
Ogpt-4o-mini-2024-07-18
gpt-4o-mini-2024-07-18Openai
40.7
9K
128K
¥1.08 / ¥4.32输入/输出
202
Ogpt-4o-2024-05-13
gpt-4o-2024-05-13Openai
40.4
14.9K
128K
¥36 / ¥108输入/输出
203
Aqwen-max-0919
qwen-max-0919Alibaba
40.1
2.4K
131K
¥2.48 / ¥9.91输入/输出
204
IB
granite-4.1-8b
granite-4.1-8bIbm
39.8
1.3K
131K
¥0.36 / ¥0.72输入/输出
205
Zglm-4-plus
glm-4-plusZai
39.5
4K
128K
¥54 / ¥54输入/输出
206
Ggemini-1.5-flash-002
gemini-1.5-flash-002Google
39.2
5.5K
2M
¥0.54 / ¥2.2输入/输出
207
Ogpt-4o-2024-08-06
gpt-4o-2024-08-06Openai
38.9
6.2K
128K
¥18 / ¥72输入/输出
208
Ogpt-4.1-nano-2025-04-14
gpt-4.1-nano-2025-04-14Openai
38.6
730
1.05M
¥14.4 / ¥57.6输入/输出
209
Aqwen2.5-72b-instruct
qwen2.5-72b-instructAlibaba
38.3
6K
131K
¥4.13 / ¥12.4输入/输出
210
Mllama-4-maverick-17b-128e-instruct
llama-4-maverick-17b-128e-instructMeta
38.0
7.1K
1M
¥1.8 / ¥6.26输入/输出
211
TEhunyuan-large-vision
hunyuan-large-visionTencent
37.7
877
-
-
212
Ggemma-3n-e4b-it
gemma-3n-e4b-itGoogle
37.4
2.9K
128K
¥0 / ¥0输入/输出
213
Xgrok-2-2024-08-13
grok-2-2024-08-13Xai
37.1
8.9K
1M
¥9 / ¥18输入/输出
214
Aclaude-3-5-sonnet-20240620
claude-3-5-sonnet-20240620Anthropic
36.8
10.6K
200K
¥21.6 / ¥108输入/输出
215
Ggemma-3-4b-it
gemma-3-4b-itGoogle
36.5
444
128K
¥1.44 / ¥1.44输入/输出
216
Ddeepseek-v2.5
deepseek-v2.5Deepseek
36.2
3.5K
1M
¥1.01 / ¥2.02输入/输出
217
Mllama-4-scout-17b-16e-instruct
llama-4-scout-17b-16e-instructMeta
35.9
5.8K
128K
¥1.44 / ¥5.62输入/输出
218
Xgrok-2-mini-2024-08-13
grok-2-mini-2024-08-13Xai
35.6
7K
1M
¥9 / ¥18输入/输出
219
Mllama-3.1-405b-instruct-bf16
llama-3.1-405b-instruct-bf16Meta
35.3
5.5K
128K
¥0 / ¥0输入/输出
220
AIolmo-3.1-32b-think
olmo-3.1-32b-thinkAllenai
35.0
2K
200K
¥14.4 / ¥57.6输入/输出
221
IAmercury
mercuryInception Ai
34.7
493
128K
¥1.8 / ¥5.4输入/输出
222
MAmistral-large-2411
mistral-large-2411Mistral
34.4
3.6K
128K
¥14.4 / ¥43.2输入/输出
223
Aclaude-3-5-haiku-20241022
claude-3-5-haiku-20241022Anthropic
34.1
11K
200K
¥5.76 / ¥28.8输入/输出
224
Mllama-3.1-405b-instruct-fp8
llama-3.1-405b-instruct-fp8Meta
33.8
8K
128K
¥0 / ¥0输入/输出
225
Aclaude-3-opus-20240229
claude-3-opus-20240229Anthropic
33.5
23.4K
200K
¥108 / ¥540输入/输出
226
MAmistral-large-2407
mistral-large-2407Mistral
33.2
5.9K
131K
¥14.4 / ¥43.2输入/输出
227
Mllama-3.3-70b-instruct
llama-3.3-70b-instructMeta
32.9
8.3K
128K
¥0 / ¥0输入/输出
228
Aamazon-nova-pro-v1.0
amazon-nova-pro-v1.0Amazon
32.6
3.4K
300K
¥5.76 / ¥23输入/输出
229
Ogpt-4-turbo-2024-04-09
gpt-4-turbo-2024-04-09Openai
32.3
11.6K
128K
¥72 / ¥216输入/输出
230
Ggemini-advanced-0514
gemini-advanced-0514Google
32.0
5.5K
-
-
231
Ggemini-1.5-flash-001
gemini-1.5-flash-001Google
31.8
7.8K
2M
¥0.54 / ¥2.2输入/输出
232
Aqwen2.5-coder-32b-instruct
qwen2.5-coder-32b-instructAlibaba
31.5
804
131K
¥2.07 / ¥6.2输入/输出
233
Ogpt-oss-20b
gpt-oss-20bOpenai
31.2
2.1K
131K
¥0.32 / ¥1.3输入/输出
234
MAmistral-small-24b-instruct-2501
mistral-small-24b-instruct-2501Mistral
30.9
1.8K
262K
¥2.88 / ¥14.4输入/输出
235
Ogpt-4-0125-preview
gpt-4-0125-previewOpenai
30.6
9.7K
8.19K
¥216 / ¥432输入/输出
236
IBibm-granite-h-small
ibm-granite-h-smallIbm
30.3
1.3K
-
-
237
Mllama-3.1-70b-instruct
llama-3.1-70b-instructMeta
30.0
7.6K
131K
¥2.88 / ¥2.88输入/输出
238
NVllama-3.1-nemotron-70b-instruct
llama-3.1-nemotron-70b-instructNvidia
29.7
1K
128K
¥0 / ¥0输入/输出
239
Ogpt-4-1106-preview
gpt-4-1106-previewOpenai
29.4
9.1K
8.19K
¥216 / ¥432输入/输出
240
Aamazon-nova-lite-v1.0
amazon-nova-lite-v1.0Amazon
29.1
2.8K
300K
¥0.43 / ¥1.73输入/输出
241
UNathene-70b-0725
-
28.8
2.1K
-
-
242
Ggemma-2-27b-it
gemma-2-27b-itGoogle
28.5
9.9K
8.19K
¥0.58 / ¥0.58输入/输出
243
COcommand-r-plus-08-2024
command-r-plus-08-2024Cohere
28.2
1.4K
128K
¥18 / ¥72输入/输出
244
COc4ai-aya-expanse-32b
c4ai-aya-expanse-32bCohere
27.9
4.2K
-
-
245
UNgemma-2-9b-it-simpo
-
27.6
841
8.19K
¥1.44 / ¥1.44输入/输出
246
NVnemotron-4-340b-instruct
nemotron-4-340b-instructNvidia
27.3
2.1K
-
-
247
TEhunyuan-standard-256k
hunyuan-standard-256kTencent
27.0
414
-
-
248
AIllama-3.1-tulu-3-70b
llama-3.1-tulu-3-70bAllenai
26.7
463
-
-
249
UNreka-core-20240904
-
26.4
951
-
-
250
Ddeepseek-coder-v2
deepseek-coder-v2Deepseek
26.1
1.8K
1M
¥1.01 / ¥2.02输入/输出
251
Ggemini-1.5-flash-8b-001
gemini-1.5-flash-8b-001Google
25.8
5.6K
2M
¥0.54 / ¥2.2输入/输出
252
M
phi-4
phi-4Microsoft
25.5
2.9K
128K
¥0.9 / ¥3.6输入/输出
253
MAministral-8b-2410
ministral-8b-2410Mistral
25.2
720
128K
¥0.72 / ¥0.72输入/输出
254
Aclaude-3-sonnet-20240229
claude-3-sonnet-20240229Anthropic
24.9
11.9K
200K
¥21.6 / ¥108输入/输出
255
Zglm-4-0520
glm-4-0520Zai
24.6
1.2K
128K
¥108 / ¥108输入/输出
256
UNjamba-1.5-large
-
24.3
1K
256K
¥0 / ¥0输入/输出
257
Aamazon-nova-micro-v1.0
amazon-nova-micro-v1.0Amazon
24.0
2.6K
128K
¥0.25 / ¥1.01输入/输出
258
NVllama-3.1-nemotron-51b-instruct
llama-3.1-nemotron-51b-instructNvidia
23.7
587
128K
¥0 / ¥0输入/输出
259
COcommand-r-plus
command-r-plusCohere
23.4
9.1K
128K
¥18 / ¥72输入/输出
260
COcommand-r-08-2024
command-r-08-2024Cohere
23.1
1.4K
128K
¥18 / ¥72输入/输出
261
Ggemma-2-9b-it
gemma-2-9b-itGoogle
22.8
6.9K
8.19K
¥1.44 / ¥1.44输入/输出
262
AIolmo-2-0325-32b-instruct
olmo-2-0325-32b-instructAllenai
22.6
261
-
-
263
Aqwen2-72b-instruct
qwen2-72b-instructAlibaba
22.3
4.6K
131K
¥4.13 / ¥12.4输入/输出
264
UNreka-flash-20240904
-
22.0
992
65.5K
¥0.72 / ¥1.44输入/输出
265
COc4ai-aya-expanse-8b
c4ai-aya-expanse-8bCohere
21.7
1.5K
-
-
266
Aclaude-3-haiku-20240307
claude-3-haiku-20240307Anthropic
21.4
14K
200K
¥1.8 / ¥9输入/输出
267
Ogpt-4-0314
gpt-4-0314Openai
21.1
4.3K
8.19K
¥216 / ¥432输入/输出
268
Ogpt-4-0613
gpt-4-0613Openai
20.8
7.5K
8.19K
¥216 / ¥432输入/输出
269
Mllama-3.1-8b-instruct
llama-3.1-8b-instructMeta
20.5
6.7K
131K
¥0.79 / ¥0.79输入/输出
270
AIllama-3.1-tulu-3-8b
llama-3.1-tulu-3-8bAllenai
20.2
454
-
-
271
Mllama-3-70b-instruct
llama-3-70b-instructMeta
19.9
17.5K
8.19K
¥3.67 / ¥5.33输入/输出
272
MAmistral-large-2402
mistral-large-2402Mistral
19.6
6.2K
262K
¥2.88 / ¥14.4输入/输出
273
Aqwq-32b-preview
qwq-32b-previewAlibaba
19.3
475
131K
¥2.07 / ¥6.2输入/输出
274
IBgranite-3.1-8b-instruct
granite-3.1-8b-instructIbm
19.0
461
-
-
275
COcommand-r
command-rCohere
18.7
6.2K
128K
¥18 / ¥72输入/输出
276
Aqwen1.5-110b-chat
qwen1.5-110b-chatAlibaba
18.4
3.3K
-
-
277
Aqwen1.5-72b-chat
qwen1.5-72b-chatAlibaba
18.1
3.5K
-
-
278
UNjamba-1.5-mini
-
17.8
1K
256K
¥0 / ¥0输入/输出
279
IBgranite-3.1-2b-instruct
granite-3.1-2b-instructIbm
17.5
489
-
-
280
MAmistral-medium
mistral-mediumMistral
17.2
2.7K
262K
¥2.88 / ¥14.4输入/输出
281
UNinternlm2_5-20b-chat
-
16.9
1.3K
-
-
282
Aqwen1.5-32b-chat
qwen1.5-32b-chatAlibaba
16.6
2.6K
-
-
283
MAmixtral-8x22b-instruct-v0.1
mixtral-8x22b-instruct-v0.1Mistral
16.3
5.9K
64K
¥14.4 / ¥43.2输入/输出
284
UNyi-1.5-34b-chat
-
16.0
2.1K
-
-
285
UNreka-flash-21b-20240226
-
15.7
2.7K
-
-
286
UNreka-flash-21b-20240226-online
-
15.4
1.7K
-
-
287
Ggemini-pro-dev-api
gemini-pro-dev-apiGoogle
15.1
1.4K
1.05M
¥14.4 / ¥86.4输入/输出
288
Ggemma-2-2b-it
gemma-2-2b-itGoogle
14.8
5.9K
128K
¥0 / ¥0输入/输出
289
Mllama-3-8b-instruct
llama-3-8b-instructMeta
14.5
11.2K
8.19K
¥0.29 / ¥0.29输入/输出
290
IBgranite-3.0-8b-instruct
granite-3.0-8b-instructIbm
14.2
603
-
-
291
Ogpt-3.5-turbo-0125
gpt-3.5-turbo-0125Openai
13.9
6.7K
16.4K
¥3.6 / ¥10.8输入/输出
292
Mphi-3-medium-4k-instruct
phi-3-medium-4k-instructMicrosoft
13.6
2.2K
4.1K
¥1.22 / ¥4.9输入/输出
293
Aqwen1.5-14b-chat
qwen1.5-14b-chatAlibaba
13.4
2.1K
-
-
294
UNdbrx-instruct-preview
-
13.1
3.5K
-
-
295
UNtulu-2-dpo-70b
-
12.8
280
-
-
296
UNzephyr-orpo-141b-A35b-v0.1
-
12.5
448
200K
¥108 / ¥432输入/输出
297
UNstarling-lm-7b-beta
-
12.2
1.7K
200K
¥5.4 / ¥18.7输入/输出
298
MAmixtral-8x7b-instruct-v0.1
mixtral-8x7b-instruct-v0.1Mistral
11.9
6.6K
32K
¥5.04 / ¥5.04输入/输出
299
Mllama-3.2-3b-instruct
llama-3.2-3b-instructMeta
11.6
1K
131K
¥0.22 / ¥0.35输入/输出
300
Mwizardlm-70b
wizardlm-70bMicrosoft
11.3
355
-
-
301
UNyi-34b-chat
-
11.0
1.1K
-
-
302
UNopenchat-3.5
-
10.7
414
-
-
303
Ddeepseek-llm-67b-chat
deepseek-llm-67b-chatDeepseek
10.4
243
1M
¥1.01 / ¥2.02输入/输出
304
Aqwen1.5-7b-chat
qwen1.5-7b-chatAlibaba
10.1
372
-
-
305
Mphi-3-small-8k-instruct
phi-3-small-8k-instructMicrosoft
9.8
2.2K
8.19K
¥1.08 / ¥4.32输入/输出
306
UNopenchat-3.5-0106
-
9.5
1.1K
-
-
307
UNstarling-lm-7b-alpha
-
9.2
666
200K
¥5.4 / ¥18.7输入/输出
308
IBgranite-3.0-2b-instruct
granite-3.0-2b-instructIbm
8.9
647
-
-
309
Ogpt-3.5-turbo-1106
gpt-3.5-turbo-1106Openai
8.6
933
16.4K
¥7.2 / ¥14.4输入/输出
310
Mllama-2-13b-chat
llama-2-13b-chatMeta
8.3
1.2K
-
-
311
Mllama-2-70b-chat
llama-2-70b-chatMeta
8.0
2.8K
-
-
312
MAmistral-7b-instruct-v0.2
mistral-7b-instruct-v0.2Mistral
7.7
1.6K
262K
¥2.88 / ¥14.4输入/输出
313
UNsmollm2-1.7b-instruct
-
7.4
328
-
-
314
Ggemma-1.1-7b-it
gemma-1.1-7b-itGoogle
7.1
2.6K
-
-
315
UNopenhermes-2.5-mistral-7b
-
6.8
270
1M
¥36 / ¥180输入/输出
316
Mwizardlm-13b
wizardlm-13bMicrosoft
6.5
203
-
-
317
Mllama-3.2-1b-instruct
llama-3.2-1b-instructMeta
6.2
1.1K
16.4K
¥0.07 / ¥0.08输入/输出
318
UNvicuna-13b
-
5.9
756
-
-
319
UNnous-hermes-2-mixtral-8x7b-dpo
-
5.6
230
1M
¥36 / ¥180输入/输出
320
Mphi-3-mini-4k-instruct
phi-3-mini-4k-instructMicrosoft
5.3
1.7K
4.1K
¥0.94 / ¥3.74输入/输出
321
UNzephyr-7b-beta
-
5.0
461
-
-
322
Mphi-3-mini-4k-instruct-june-2024
phi-3-mini-4k-instruct-june-2024Microsoft
4.7
1K
4.1K
¥0.94 / ¥3.74输入/输出
323
UNvicuna-7b
-
4.5
162
-
-
324
UNvicuna-33b
-
4.2
1K
-
-
325
Aqwen-14b-chat
qwen-14b-chatAlibaba
3.9
214
32.8K
¥1.04 / ¥3.1输入/输出
326
Ggemma-7b-it
gemma-7b-itGoogle
3.6
676
-
-
327
Mcodellama-34b-instruct
codellama-34b-instructMeta
3.3
320
-
-
328
Gpalm-2
palm-2Google
3.0
251
-
-
329
Ggemma-1.1-2b-it
gemma-1.1-2b-itGoogle
2.7
1.2K
-
-
330
MAmistral-7b-instruct
mistral-7b-instructMistral
2.4
448
262K
¥2.88 / ¥14.4输入/输出
331
UNsnowflake-arctic-instruct
-
2.1
2.8K
-
-
332
Mllama-2-7b-chat
llama-2-7b-chatMeta
1.8
835
128K
¥4.03 / ¥48输入/输出
333
NVllama2-70b-steerlm-chat
llama2-70b-steerlm-chatNvidia
1.5
208
-
-
334
Mphi-3-mini-128k-instruct
phi-3-mini-128k-instructMicrosoft
1.2
2.1K
128K
¥0.94 / ¥3.74输入/输出
335
Aqwen1.5-4b-chat
qwen1.5-4b-chatAlibaba
0.9
692
-
-
336
UNstripedhyena-nous-7b
-
0.6
334
-
-
337
Ggemma-2b-it
gemma-2b-itGoogle
0.3
377
-
-
338
UNchatglm3-6b
-
0.0
205
200K
¥5.4 / ¥18.7输入/输出
claude-opus-4-6-thinking 为什么排第一
claude-opus-4-6-thinking 当前排名第 1，厂商为 anthropic，百分制分数 100.0，样本量 11.9K。如果你的任务与本榜单口径一致，可以优先把它作为候选模型，再结合价格、上下文长度和可用性做最终选择。
不要只看第一名
先选择最接近真实任务的排行榜，再比较 Top 模型的分数和样本量。分数越高说明同一榜单内偏好表现越强；样本量越高通常结果更稳定。最后再结合价格、上下文长度、开源闭源和厂商可用性筛选。
继续比较相近能力
长程任务常见问题
长程任务排行榜看什么指标？
主要看排名、百分制分数、样本量和来源。分数用于快速比较同一榜单内模型表现，样本量用于判断结果稳定性。
为什么不同榜单不能直接混合成总分？
不同榜单的任务、样本和评测口径不同，模力榜默认只在同一榜单内排序，避免把写作、代码、图像等能力强行合并。
长程任务模型应该怎么选？
优先看与你任务最接近的榜单，再结合价格、上下文长度、开源闭源和厂商可用性。排名靠前不代表适合所有预算和部署方式。
榜单多久更新？
页面展示的是最新成功采集的公开榜单数据。当前优先使用 LMArena leaderboard dataset，并在页面来源中保留原始链接。