AnyInt加速 AGI
Rankings

驅動 AnyInt 的性能領先者視覺脈搏。

篩選跨越智能、速度、延遲、成本、上下文和企業級就緒性的基準，為您的 AI 工作流程部署合適的智能。

Models tracked30
Providers18
Top intelligence57
Fastest output301 tok/s
Intelligence

智能指數排名

AnyInt 智能指數 v4.0 結合了 10 項評估，包括推理、知識、數學和編程基準。分數越高表示整體智能越好。

01

Gemini 3.1 Pro Preview

Google
Score
57
02

GPT-5.3 Codex (xhigh)

OpenAI
Score
54
03

Claude Opus 4.6 (max)

Anthropic
Score
53
04

Claude Sonnet 4.6 (max)

Anthropic
Score
52
05

GPT-5.2 (xhigh)

OpenAI
Score
51
06

GLM-5

Z AI
Score
50
07

GPT-5.2 Codex (xhigh)

OpenAI
Score
49
08

Kimi K2.5

Kimi
Score
47
09

Claude Opus 4.6

Anthropic
Score
46
10

Gemini 3 Flash

Google
Score
46
11

Qwen3.5 397B A17B

Alibaba
Score
45
12

MiniMax-M2.5

MiniMax
Score
42
13

DeepSeek V3.2

DeepSeek
Score
42
14

Grok 4

xAI
Score
42
15

MiMo-V2-Flash (Feb 2026)

Xiaomi
Score
41
16

Grok 4.1 Fast

xAI
Score
39
17

Claude Haiku 4.5

Anthropic
Score
37
18

KAT-Coder-Pro V1

KwaiKAT
Score
36
19

Nova 2.0 Pro (medium)

AWS
Score
36
20

gpt-oss-120B (high)

OpenAI(?)
Score
33
21

K-EXAONE

LG AI Research
Score
32
22

gpt-oss-20B (high)

OpenAI(?)
Score
24
23

NVIDIA Nemotron Nano 3

NVIDIA
Score
24
24

K2 Think V2

Kimi(?)
Score
24
25

Mistral Large 3

Mistral
Score
23
26

Llama 4 Maverick

Meta
Score
18
Context

上下文視窗排名

上下文視窗大小決定了模型在單次請求中能處理多少文字。更大的上下文視窗能分析更長的文件和更複雜的對話。

01

Grok 4.1 Fast

xAI
Window
2M
02

Llama 4 Maverick

Meta
Window
1M
03

Gemini 3.1 Pro Preview

Google
Window
1M
04

NVIDIA Nemotron Nano 4

NVIDIA
Window
1M
05

Gemini 3

Google
Window
1M
06

GPT-5.3 (xhigh)

OpenAI
Window
1M
07

GPT-5.2 Codex (xhigh)

OpenAI
Window
1M
08

K2 Think V2

Kimi
Window
400K
09

Qwen3.5 397B A17B

Alibaba
Window
400K
10

Mistral Large 3

Mistral
Window
400K
11

Grok 4

xAI
Window
262K
12

Nova 2.0 Pro (medium)

AWS
Window
256K
13

gpt-oss-120B (high)

OpenAI
Window
256K
14

Kimi K2.5

Kimi
Window
256K
15

K-EXAONE

LG AI Research
Window
256K
16

gpt-oss-20B (high)

OpenAI
Window
256K
17

MiMo-V2-Flash (Feb 2026)

Xiaomi
Window
256K
18

KAT-Coder-Pro V1

KwaiKAT
Window
256K
19

MiniMax-M2.5

MiniMax
Window
205K
20

Claude Opus 4.6 (max)

Anthropic
Window
200K
21

Claude Opus 4.6

Anthropic
Window
200K
22

Claude Sonnet 4.6 (max)

Anthropic
Window
200K
23

Claude Haiku 4.5

Anthropic
Window
200K
24

GLM-5

Z AI
Window
131K
25

DeepSeek V3.2

DeepSeek
Window
128K
Performance

輸出速度排名

輸出速度衡量模型生成 token 的速度。更高的速度能實現更靈敏的互動和更快的長文本完成。

01

gpt-oss-120B (high)

OpenAI
Speed
301 tok/s
02

gpt-oss-20B (high)

OpenAI
Speed
299 tok/s
03

Grok 4.1 Fast

xAI
Speed
239 tok/s
04

Gemini 3 Flash

Google
Speed
213 tok/s
05

MiMo-V2-Flash (Feb 2026)

Xiaomi
Speed
158 tok/s
06

NVIDIA Nemotron Nano 3

NVIDIA
Speed
154 tok/s
07

Nova 2.0 Pro (medium)

AWS
Speed
132 tok/s
08

Claude Haiku 4.5

Anthropic
Speed
124 tok/s
09

Llama 4 Maverick

Meta
Speed
113 tok/s
10

GPT-5.3 Codex (xhigh)

OpenAI
Speed
98 tok/s
11

GPT-5.2 (xhigh)

OpenAI
Speed
91 tok/s
12

Gemini 3.1 Pro Preview

Google
Speed
88 tok/s
13

GPT-5.2 Codex (xhigh)

OpenAI
Speed
87 tok/s
14

Qwen3.5 397B A17B

Alibaba
Speed
81 tok/s
15

Claude Opus 4.6 (max)

Anthropic
Speed
72 tok/s
16

Claude 4.6

Anthropic
Speed
69 tok/s
17

Claude Opus 4.6

Anthropic
Speed
68 tok/s
18

GLM-5

Z AI
Speed
57 tok/s
19

KAT-Coder-Pro V1

KwaiKAT
Speed
56 tok/s
20

Claude Sonnet 4.6 (max)

Anthropic
Speed
54 tok/s
21

MiniMax-M2.5

MiniMax
Speed
50 tok/s
22

Mistral Large 3

Mistral
Speed
46 tok/s
23

DeepSeek V3.2

DeepSeek
Speed
42 tok/s
24

Kimi K2.5

Kimi
Speed
41 tok/s
25

Grok 4

xAI
Speed
41 tok/s