Google:模型智能、性能与价格

Google
Google

分析 Google 各模型的关键指标,包括质量、价格、输出速度、延迟、上下文窗口等。 本分析旨在帮助你根据使用场景,选择 Google 提供的最佳模型。

最智能

#1
Claude Opus 5 (max)
Claude Opus 5 (max)
61
#2
Claude Opus 5 (xhigh)
Claude Opus 5 (xhigh)
60
#3
Claude Fable 5 (with fallback)
Claude Fable 5 (with fallback)
60
#4
Claude Opus 5 (high)
Claude Opus 5 (high)
59
#5
Claude Opus 5 (medium)
Claude Opus 5 (medium)
56

Intelligence Index

共 57 个模型

速度最快

#1
gpt-oss-120b (high) Vertex
gpt-oss-120b (high) Vertex
423 t/s
#2
gpt-oss-120b (low) Vertex
gpt-oss-120b (low) Vertex
408 t/s
#3
Gemini 3.5 Flash-Lite AI Studio
Gemini 3.5 Flash-Lite AI Studio
346 t/s
#4
Gemini 2.5 Flash-Lite (AI Studio)
Gemini 2.5 Flash-Lite (AI Studio)
337 t/s
#5
Gemini 3.1 Flash-Lite (AI Studio)
Gemini 3.1 Flash-Lite (AI Studio)
312 t/s

输出速度

共 57 个模型

价格最低

#1
Gemini 2.5 Flash-Lite (AI Studio)
Gemini 2.5 Flash-Lite (AI Studio)
$0.07
#2
Gemini 2.5 Flash-Lite (AI Studio)
Gemini 2.5 Flash-Lite (AI Studio)
$0.07
#3
gpt-oss-20b (high) Vertex
gpt-oss-20b (high) Vertex
$0.09
#4
gpt-oss-20b (low) Vertex
gpt-oss-20b (low) Vertex
$0.09
#5
gpt-oss-120b (high) Vertex
gpt-oss-120b (high) Vertex
$0.12

每 100 万 token 的混合价格

共 57 个模型

表示推理模型

Google 提供 57 个模型,每个模型的智能、性能和价格特征各不相同。 下方对比了各模型的关键指标。

  • 智能方面,Google 上表现最好的模型是 Claude Opus 5 (max)(61)、Claude Opus 5 (xhigh)(60)和Claude Fable 5 (with fallback)(60)。
  • 输出速度方面,最快的模型是 gpt-oss-120b (high) Vertex(423 t/s)、gpt-oss-120b (low) Vertex(408 t/s)和Gemini 3.5 Flash-Lite AI Studio(346 t/s)。
  • 延迟方面,Gemini 2.5 Flash-Lite (AI Studio)(0.33 秒)、Gemini 2.5 Flash (AI Studio)(0.47 秒)和Gemini 2.5 Flash (Vertex)(0.56 秒) 的首个答案 Token 延迟最低。
  • 价格方面,Gemini 2.5 Flash-Lite (AI Studio)($0.07)、Gemini 2.5 Flash-Lite (AI Studio)($0.07)和gpt-oss-20b (high) Vertex($0.09) 每 100 万 token 的混合价格最低。
  • 上下文窗口方面,Llama 4 Scout Vertex(1M)、Claude Opus 5 (max)(1M)和Claude Opus 5 (xhigh)(1M) 支持 Google 上最大的上下文窗口。

亮点

Artificial Analysis Intelligence Index · Higher is better
Output tokens per second · Higher is better
USD per 1M tokens (blended) · Lower is better

智能评测

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR
Reasoning models are indicated by a lightbulb icon

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Intelligence Evaluations

Intelligence evaluations measured independently by Artificial Analysis · Higher is better

Agentic real-world work tasks, (Elo-500)/2000

Agentic tool use

Agentic coding & terminal use

Coding

Reasoning & knowledge

Scientific reasoning

Physics reasoning

Long context reasoning

Agentic knowledge work, Elo

Agentic SaaS workflows

Legal agentic work, criterion pass rate

Agentic business operations

Instruction following

Long-horizon agentic tasks

Kubernetes incident root-cause analysis

Visual reasoning

Reasoning models are indicated by a lightbulb icon

While model intelligence generally translates across use cases, specific evaluations may be more relevant for certain use cases.

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Intelligence Index vs. Price

Blended at 7:2:1 (cache-input-output) · USD per 1M tokens (blended)
Most attractive quadrant
Pareto line
Reasoning models are indicated by a lightbulb icon

While higher intelligence models are typically more expensive, they do not all follow the same price-quality curve.

上下文窗口

Context Window

Context window: tokens limit · Higher is better
Reasoning models are indicated by a lightbulb icon

Larger context windows are relevant to RAG (Retrieval Augmented Generation) LLM workflows which typically involve reasoning and information retrieval of large amounts of data.

Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

价格

Intelligence Index vs. Price

Blended at 7:2:1 (cache-input-output) · USD per 1M tokens (blended)
Most attractive quadrant
Pareto line
Reasoning models are indicated by a lightbulb icon

While higher intelligence models are typically more expensive, they do not all follow the same price-quality curve.

性能摘要

Output Speed vs. Price

Output speed: output tokens per second · USD per 1M tokens (blended)
Most attractive quadrant
Pareto line
Reasoning models are indicated by a lightbulb icon

Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).

速度

按输出速度(每秒 token 数)衡量

Output Speed

Output tokens per second · Higher is better
Reasoning models are indicated by a lightbulb icon

Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

延迟

按首 Token 延迟(秒)衡量

Latency: Time To First Answer Token

Seconds to first answer token received · Accounts for reasoning model 'thinking' time
Reasoning models are indicated by a lightbulb icon

Time to first answer token received, in seconds, after API request sent. For reasoning models, this includes the 'thinking' time of the model before providing an answer. For models which do not support streaming, this represents time to receive the completion.

端到端响应时间

Seconds to output 500 tokens, calculated based on time to first token, 'thinking' time for reasoning models, and output speed

End-to-End Response Time vs. Price

End-to-end response time: end-to-end seconds to output 500 tokens · USD per 1M tokens (blended)
Most attractive quadrant
Reasoning models are indicated by a lightbulb icon

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).

进一步分析
Anthropic 标志
Claude Opus 5 (max)
1M
专有
61
$3.31
54
66.17
75.40
--
Anthropic 标志
Claude Opus 5 (xhigh)
1M
专有
60
$2.52
55
31.72
40.81
--
Anthropic 标志
Claude Fable 5 (with fallback)
1M
专有
60
$3.33
71
163.01
170.02
--
Anthropic 标志
Claude Opus 5 (high)
1M
专有
59
$1.67
55
20.98
30.06
--
Anthropic 标志
Claude Opus 5 (medium)
1M
专有
56
$0.95
53
4.62
14.02
--
Anthropic 标志
Claude Opus 4.8 (max)
1M
专有
56
$2.58
66
28.04
35.59
--
Anthropic 标志
Claude Opus 4.7 (max)
1M
专有
54
$3.84
67
15.00
22.46
--
Anthropic 标志
Claude Sonnet 5 (max)
1M
专有
53
$2.22
83
213.41
219.42
--
Anthropic 标志
Claude Opus 5 (low)
1M
专有
51
$0.55
52
2.71
12.29
--
Google 标志
Gemini 3.5 Flash AI Studio
1M
专有
50
$0.69
179
24.97
27.77
--
Google 标志
Gemini 3.6 Flash AI Studio
1M
专有
50
$0.56
200
15.67
18.16
--
Anthropic 标志
Claude Sonnet 4.6 (max)
200k
专有
47
$1.62
61
128.32
136.54
--
Google 标志
Gemini 3.1 Pro Preview (AI Studio)
1M
专有
46
$0.34
126
32.39
36.35
--
Google 标志
Gemini 3.1 Pro Preview (Vertex)
1M
专有
46
$0.43
138
29.23
32.86
--
Google 标志
Gemini 3.5 Flash (medium)
1M
专有
45*
--
182
17.04
19.78
--
Anthropic 标志
Claude Opus 4.6 (max)
1M
专有
44*
--
46
17.16
28.11
--
Anthropic 标志
Claude Opus 4.5 Vertex
200k
专有
41*
--
51
13.19
23.06
--
Google 标志
Gemini 3 Pro Preview (high) (Vertex)
1M
专有
40*
--
--
--
--
--
Z AI 标志
GLM-5
200k
开放
40*
--
183
0.99
20.68
16.96
Anthropic 标志
Claude Opus 4.6 (high)
1M
专有
38*
--
40
1.94
14.51
--
Google 标志
Gemini 3 Flash (AI Studio)
1M
专有
38*
--
178
7.31
10.12
--
Google 标志
Gemini 3.5 Flash-Lite AI Studio
1M
专有
36
$0.10
346
10.30
11.75
--
Anthropic 标志
Claude 4.5 Sonnet Vertex
1M
专有
36
$0.44
44
11.28
22.71
--
Anthropic 标志
Claude Sonnet 4.6 (Non-reasoning)
200k
专有
36*
--
45
1.22
12.41
--
Google 标志
Gemini 3.5 Flash (minimal) AI Studio
1M
专有
35*
--
171
1.05
3.97
--
Anthropic 标志
Claude Opus 4.5 Vertex
200k
专有
35*
--
44
1.03
12.37
--
Anthropic 标志
Claude 4.1 Opus Vertex
200k
专有
34*
--
--
--
--
--
Z AI 标志
GLM-4.7
200k
开放
34
$0.35
129
0.70
20.09
15.51
Google 标志
Gemini 3 Pro Preview (low) (Vertex)
1M
专有
33*
--
--
--
--
--
Kimi 标志
Kimi K2 Thinking Vertex
262k
开放
33*
--
148
0.85
17.73
13.51
Anthropic 标志
Claude 4 Opus Vertex
200k
专有
31*
--
--
--
--
--
Anthropic 标志
Claude 4.5 Haiku Vertex
200k
专有
30
$0.24
100
16.63
21.63
--
Google 标志
Gemma 4 31B (AI Studio)
262k
开放
29
$0.00
35
1.10
64.26
49.03
Anthropic 标志
Claude 4.5 Sonnet Vertex
1M
专有
29*
--
39
1.10
13.99
--
Anthropic 标志
Claude 4 Sonnet Vertex
1M
专有
29
$0.46
--
--
--
--
MiniMax 标志
MiniMax-M2 Vertex
197k
开放
28*
--
136
0.65
18.97
14.66
Anthropic 标志
Claude 4.1 Opus Vertex
200k
专有
28*
--
--
--
--
--
Google 标志
Gemini 3 Flash (AI Studio)
1M
专有
27*
--
181
0.87
3.63
--
Z AI 标志
GLM-4.7
200k
开放
27*
--
126
0.68
4.65
--
Google 标志
Gemini 2.5 Pro Vertex
1M
专有
26
$0.22
113
26.81
31.25
--
Google 标志
Gemini 2.5 Pro (AI Studio)
1M
专有
26
$0.22
136
22.36
26.03
--
Google 标志
Gemma 4 26B A4B AI Studio
262k
开放
26
$0.00
47
1.07
54.26
42.55
Anthropic 标志
Claude 4 Opus Vertex
200k
专有
25*
--
--
--
--
--
Anthropic 标志
Claude 4 Sonnet Vertex
1M
专有
25*
--
--
--
--
--
Google 标志
Gemini 3.1 Flash-Lite (AI Studio)
1M
专有
25
$0.04
312
6.13
7.73
--
OpenAI 标志
gpt-oss-120b (high) Vertex
131k
开放
24
$0.05
423
0.39
6.30
4.72
Anthropic 标志
Claude 4.5 Haiku Vertex
200k
专有
24*
--
92
0.66
6.09
--
Anthropic 标志
Claude 3.7 Sonnet Vertex
200k
专有
24*
--
--
--
--
--
Google 标志
Gemini 2.5 Pro (May) (AI Studio)
1M
专有
22*
--
--
--
--
--
DeepSeek 标志
DeepSeek V3.1 Vertex
164k
开放
21*
--
129
0.83
4.69
--
DeepSeek 标志
DeepSeek V3.1 Vertex
164k
开放
21*
--
79
0.84
32.60
25.40
DeepSeek 标志
DeepSeek R1 0528 Vertex
164k
开放
20*
--
151
1.05
17.56
13.20
Google 标志
Gemini 2.5 Flash (AI Studio)
1M
专有
20*
--
218
13.78
16.07
--
Google 标志
Gemini 2.5 Flash (Vertex)
1M
专有
20*
--
179
25.04
27.84
--
Alibaba 标志
Qwen3 235B 2507 Vertex
262k
开放
18*
--
83
0.84
6.83
--
Alibaba 标志
Qwen3 Coder 480B Vertex
262k
开放
18*
--
151
1.73
5.04
--
Alibaba 标志
Qwen3 Next 80B A3B Vertex
262k
开放
17
$0.04
160
0.77
16.42
12.52
Google 标志
Gemini 2.5 Flash-Lite (Sep) (AI Studio)
1M
专有
15*
--
--
--
--
--
OpenAI 标志
gpt-oss-120b (low) Vertex
131k
开放
15
$0.01
408
0.38
6.52
4.91
OpenAI 标志
gpt-oss-20b (high) Vertex
131k
开放
15
$0.03
312
3.32
11.35
6.42
OpenAI 标志
gpt-oss-20b (low) Vertex
131k
开放
14*
--
246
0.41
10.56
8.12
Google 标志
Gemini 2.5 Flash (Vertex)
1M
专有
14*
--
176
0.56
3.41
--
Google 标志
Gemini 2.5 Flash (AI Studio)
1M
专有
14*
--
205
0.47
2.90
--
Alibaba 标志
Qwen3 Next 80B A3B Vertex
262k
开放
14*
--
184
0.75
3.47
--
Google 标志
Gemini 2.5 Flash-Lite (Sep) (AI Studio)
1M
专有
13*
--
--
--
--
--
Google 标志
Gemini 2.5 Flash-Lite (AI Studio)
1M
专有
11*
--
337
26.05
27.53
--
Google 标志
Gemini 2.0 Flash (exp) (AI Studio)
1M
专有
11*
--
--
--
--
--
Meta 标志
Llama 4 Scout Vertex
1.31M
开放
10
$0.01
136
0.83
4.52
--
Meta 标志
Llama 3.3 70B Vertex
128k
开放
9
$0.11
129
0.69
4.55
--
Google 标志
Gemma 3 27B (AI Studio)
128k
开放
7
$0.00
--
--
--
--
Google 标志
Gemini 2.5 Flash-Lite (AI Studio)
1M
专有
7*
--
232
0.33
2.49
--
Google 标志
Gemma 3 12B (AI Studio)
128k
开放
6
$0.00
--
--
--
--
Google 标志
Gemma 3 4B (AI Studio)
128k
开放
1*
--
--
--
--
--
Google 标志
Gemma 3n E2B (AI Studio)
32k
开放
1*
--
--
--
--
--
Google 标志
Gemma 3 1B (AI Studio)
32k
开放
1*
--
--
--
--
--

关键定义

Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

常见问题

关于 Google 的常见问题

Google 提供我们跟踪的 57 个模型:Claude Opus 5 (max)Claude Opus 5 (xhigh)Claude Fable 5 (with fallback)Claude Opus 5 (high)Claude Opus 5 (medium)Claude Opus 4.8 (max)Claude Opus 4.7 (max)Claude Sonnet 5 (max)Claude Opus 5 (low)Gemini 3.5 Flash AI StudioGemini 3.6 Flash AI StudioClaude Sonnet 4.6 (max)Gemini 3.1 Pro Preview (AI Studio)Gemini 3.1 Pro Preview (Vertex)Gemini 3.5 Flash (medium)Claude Opus 4.6 (max)Claude Opus 4.5 VertexGLM-5Claude Opus 4.6 (high)Gemini 3 Flash (AI Studio)Gemini 3.5 Flash-Lite AI StudioClaude 4.5 Sonnet VertexClaude Sonnet 4.6 (Non-reasoning)Gemini 3.5 Flash (minimal) AI StudioClaude Opus 4.5 VertexGLM-4.7Kimi K2 Thinking VertexClaude 4.5 Haiku VertexGemma 4 31B (AI Studio)Claude 4.5 Sonnet VertexMiniMax-M2 VertexGemini 3 Flash (AI Studio)GLM-4.7Gemini 2.5 Pro VertexGemini 2.5 Pro (AI Studio)Gemma 4 26B A4B AI StudioGemini 3.1 Flash-Lite (AI Studio)gpt-oss-120b (high) VertexClaude 4.5 Haiku VertexDeepSeek V3.1 VertexDeepSeek V3.1 VertexDeepSeek R1 0528 VertexGemini 2.5 Flash (AI Studio)Gemini 2.5 Flash (Vertex)Qwen3 235B 2507 VertexQwen3 Coder 480B VertexQwen3 Next 80B A3B Vertexgpt-oss-120b (low) Vertexgpt-oss-20b (high) Vertexgpt-oss-20b (low) VertexGemini 2.5 Flash (Vertex)Gemini 2.5 Flash (AI Studio)Qwen3 Next 80B A3B VertexGemini 2.5 Flash-Lite (AI Studio)Llama 4 Scout VertexLlama 3.3 70B VertexGemini 2.5 Flash-Lite (AI Studio)

Google 上最智能的模型是 Claude Opus 5 (max),Intelligence Index 得分为 61。

按输出速度计算,Google 上最快的模型是 gpt-oss-120b (high) Vertex,速度为每秒 423.4 个 token。

Google 上首个答案 Token 延迟最低的模型是 Gemini 2.5 Flash-Lite (AI Studio),延迟为 0.33 秒。延迟越低,初始响应越快。

按混合价格计算,Google 上最实惠的模型是 Gemini 2.5 Flash-Lite (AI Studio),每 100 万 token 的价格为 $0.07(缓存命中/输入/输出比例为 7:2:1)。

Google 上各模型价格最多相差 115 倍,从最实惠的 Gemini 2.5 Flash-Lite (AI Studio)(每 100 万 token $0.07)到最昂贵的 Claude Fable 5 (with fallback)(每 100 万 token $7.70)。

是,Google 提供兼容 OpenAI 的 API,可以轻松从 OpenAI 切换,或继续使用现有的 OpenAI SDK 集成。

Google 上 57 个模型中,有 44 个支持用于结构化输出的 JSON 模式。

是,Google 上全部 57 个模型均支持函数调用(工具使用)。

会。服务商性能可能因基础设施变化、负载均衡和更新而随时间波动。我们持续对所有服务商进行基准测试,并在“随时间变化”图表中展示历史性能趋势。

选择 Google 上的模型时,请考虑:智能(适合质量敏感型任务)、输出速度(适合吞吐量密集型任务)、延迟(适合需要快速首次响应的交互式应用)、价格(适合成本敏感型工作负载),以及上下文窗口大小、JSON 模式或函数调用支持等功能。