OpenAI:模型智能、性能与价格

OpenAI
OpenAI

分析 OpenAI 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 OpenAI 提供的最佳模型。

最智能

#1
GPT-5.6 Sol (max)
GPT-5.6 Sol (max)
59
#2
GPT-5.6 Sol (xhigh)
GPT-5.6 Sol (xhigh)
58
#3
GPT-5.6 Sol (high)
GPT-5.6 Sol (high)
56
#4
GPT-5.6 Terra (max)
GPT-5.6 Terra (max)
55
#5
GPT-5.5 (xhigh)
GPT-5.5 (xhigh)
55

Intelligence Index

共 61 个模型

速度最快

#1
o3-mini
o3-mini
211 t/s
#2
o3-mini (high)
o3-mini (high)
201 t/s
#3
GPT-4.1 nano
GPT-4.1 nano
187 t/s
#4
GPT-5.4 mini (xhigh)
GPT-5.4 mini (xhigh)
180 t/s
#5
GPT-5.6 Luna (max)
GPT-5.6 Luna (max)
178 t/s

输出速度

共 61 个模型

价格最低

#1
GPT-5 nano (high)
GPT-5 nano (high)
$0.05
#2
GPT-5 nano (medium)
GPT-5 nano (medium)
$0.05
#3
GPT-5 nano (minimal)
GPT-5 nano (minimal)
$0.05
#4
GPT-4.1 nano
GPT-4.1 nano
$0.08
#5
GPT-4o mini
GPT-4o mini
$0.14

每 100 万 token 的混合价格

共 61 个模型

表示推理模型

OpenAI 提供 61 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,OpenAI 上表现最好的模型是 GPT-5.6 Sol (max)(59)、GPT-5.6 Sol (xhigh)(58)和GPT-5.6 Sol (high)(56)。
  • 输出速度方面,最快的模型是 o3-mini(211 t/s)、o3-mini (high)(201 t/s)和GPT-4.1 nano(187 t/s)。
  • 延迟方面,GPT-5.4 nano(0.58 秒)、GPT-4.1 nano(0.62 秒)和GPT-5.4 mini(0.67 秒) 的首个答案 Token 延迟最低。
  • 价格方面,GPT-5 nano (high)($0.05)、GPT-5 nano (medium)($0.05)和GPT-5 nano (minimal)($0.05) 每 100 万 token 的混合价格最低。 各模型价格最多相差 2.7 倍。
  • 上下文窗口方面,GPT-5.5 (xhigh)(1M)、GPT-5.5 (high)(1M)和GPT-5.4 (xhigh)(1M) 支持 OpenAI 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
Pareto line
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
OpenAI 标志
GPT-5.6 Sol (max)
1M
专有
59
$1.23
66
139.09
146.69
--
OpenAI 标志
GPT-5.6 Sol (xhigh)
1M
专有
58
$0.83
65
48.05
55.79
--
OpenAI 标志
GPT-5.6 Sol (high)
1M
专有
56
$0.55
62
12.87
20.99
--
OpenAI 标志
GPT-5.6 Terra (max)
1M
专有
55
$0.51
122
171.60
175.69
--
OpenAI 标志
GPT-5.5 (xhigh)
1.05M
专有
55
$1.17
79
79.27
85.56
--
OpenAI 标志
GPT-5.6 Sol (medium)
1M
专有
54
$0.39
61
7.89
16.09
--
OpenAI 标志
GPT-5.5 (high)
1.05M
专有
53
$0.80
72
36.19
43.10
--
OpenAI 标志
GPT-5.6 Terra (xhigh)
1M
专有
52
$0.31
116
25.28
29.59
--
OpenAI 标志
GPT-5.4 (xhigh)
1.05M
专有
51
$1.11
133
135.88
139.63
--
OpenAI 标志
GPT-5.6 Luna (max)
1M
专有
51
$0.05
178
135.99
138.80
--
OpenAI 标志
GPT-5.5 (medium)
1.05M
专有
50
$0.50
73
9.40
16.26
--
OpenAI 标志
GPT-5.6 Sol (low)
1M
专有
49
$0.24
61
3.74
11.98
--
OpenAI 标志
GPT-5.6 Luna (xhigh)
1M
专有
49
$0.03
168
46.39
49.37
--
OpenAI 标志
GPT-5.6 Terra (high)
1M
专有
49
$0.22
116
3.33
7.65
--
OpenAI 标志
GPT-5.6 Luna (high)
1M
专有
46
$0.02
168
11.71
14.68
--
OpenAI 标志
GPT-5.6 Terra (medium)
1M
专有
46
$0.12
113
1.83
6.26
--
OpenAI 标志
GPT-5.3 Codex (xhigh)
400k
专有
44*
--
124
84.77
88.80
--
OpenAI 标志
GPT-5.5 (low)
1.05M
专有
43
$0.26
61
2.07
10.22
--
OpenAI 标志
GPT-5.2 (xhigh)
400k
专有
42*
--
82
168.27
174.34
--
OpenAI 标志
GPT-5.6 Sol (Non-reasoning)
1M
专有
41
$0.24
60
1.05
9.38
--
OpenAI 标志
GPT-5.6 Terra (low)
1M
专有
40
$0.09
112
1.64
6.10
--
OpenAI 标志
GPT-5.4 mini (xhigh)
400k
专有
40
$0.49
180
9.59
12.36
--
OpenAI 标志
GPT-5.4 (low)
1.05M
专有
39*
--
107
1.65
6.33
--
OpenAI 标志
GPT-5.4 nano (xhigh)
400k
专有
38
$0.15
161
4.84
7.94
--
OpenAI 标志
GPT-5.6 Luna (medium)
1M
专有
38
$0.01
151
3.01
6.33
--
OpenAI 标志
GPT-5.2 (medium)
400k
专有
38*
--
64
8.21
16.04
--
OpenAI 标志
GPT-5.1 (high)
272k
专有
37
$0.30
105
53.96
58.73
--
OpenAI 标志
GPT-5.5 (Non-reasoning)
1.05M
专有
35
$0.21
66
0.92
8.50
--
OpenAI 标志
GPT-5 (high)
400k
专有
35
$0.26
91
94.35
99.87
--
OpenAI 标志
GPT-5.6 Terra (Non-reasoning)
1M
专有
34
$0.10
107
0.79
5.48
--
OpenAI 标志
GPT-5 (medium)
400k
专有
34*
--
80
34.60
40.87
--
OpenAI 标志
GPT-5.5 Instant (May 2026)
400k
专有
34*
--
--
--
--
--
OpenAI 标志
GPT-5.6 Luna (low)
1M
专有
33
$0.01
137
2.12
5.77
--
OpenAI 标志
o3-pro
200k
专有
33*
--
--
--
--
--
OpenAI 标志
GPT-5 (low)
400k
专有
31*
--
75
11.36
18.04
--
OpenAI 标志
GPT-5 mini (medium)
400k
专有
31*
--
95
16.71
21.96
--
OpenAI 标志
o3
200k
专有
30*
--
--
--
--
--
OpenAI 标志
o3
200k
专有
30*
--
120
8.41
12.57
--
OpenAI 标志
GPT-5.4 nano
400k
专有
30*
--
164
6.21
9.26
--
OpenAI 标志
GPT-5.4 mini (medium)
272k
专有
30*
--
164
12.78
15.82
--
OpenAI 标志
GPT-5.5 Instant (June 2026)
400k
专有
29
$0.54
168
1.00
15.93
11.94
OpenAI 标志
GPT-5.4 (Non-reasoning)
1.05M
专有
28*
--
95
0.78
6.06
--
OpenAI 标志
GPT-5.6 Luna (Non-reasoning)
1M
专有
27
$0.01
162
0.82
3.91
--
OpenAI 标志
GPT-5.2
400k
专有
26*
--
65
0.99
8.73
--
OpenAI 标志
o4-mini (high)
200k
专有
26*
--
--
--
--
--
OpenAI 标志
o4-mini (high)
200k
专有
26*
--
170
25.54
28.48
--
OpenAI 标志
GPT-5 mini (high)
400k
专有
25
$0.04
88
101.60
107.26
--
OpenAI 标志
o1
200k
专有
23*
--
--
--
--
--
OpenAI 标志
GPT-5.1
400k
专有
20*
--
103
1.03
5.89
--
OpenAI 标志
GPT-5 nano (high)
400k
专有
20*
--
141
119.99
123.53
--
OpenAI 标志
GPT-4.1
1M
专有
19*
--
129
1.05
4.91
--
OpenAI 标志
GPT-5 nano (medium)
400k
专有
19*
--
146
48.38
51.79
--
OpenAI 标志
o3-mini
200k
专有
19*
--
211
6.09
8.47
--
OpenAI 标志
o1-pro
200k
专有
19*
--
--
--
--
--
OpenAI 标志
GPT-5.4 nano
400k
专有
18*
--
165
0.58
3.60
--
OpenAI 标志
GPT-5 (minimal)
400k
专有
17*
--
60
1.23
9.56
--
OpenAI 标志
GPT-5 (minimal) private
400k
专有
17*
--
--
--
--
--
OpenAI 标志
GPT-5.4 mini
272k
专有
17*
--
168
0.67
3.64
--
OpenAI 标志
o3-mini (high)
200k
专有
16
--
201
23.85
26.34
--
OpenAI 标志
GPT-4.1 mini
1M
专有
15
$0.07
72
0.83
7.79
--
OpenAI 标志
GPT-5 mini (minimal)
400k
专有
14*
--
89
0.89
6.49
--
OpenAI 标志
GPT-5 mini (minimal) private
400k
专有
14*
--
--
--
--
--
OpenAI 标志
GPT-4o (Nov)
128k
专有
11*
--
150
0.85
4.20
--
OpenAI 标志
GPT-4o (Aug)
128k
专有
10*
--
85
1.19
7.04
--
OpenAI 标志
GPT-4o (Aug)
128k
专有
10*
--
--
--
--
--
OpenAI 标志
GPT-4.1 nano
1M
专有
10
$0.04
187
0.62
3.29
--
OpenAI 标志
GPT-4o (May)
128k
专有
9*
--
80
1.25
7.51
--
OpenAI 标志
GPT-5 nano (minimal)
400k
专有
8*
--
145
0.87
4.33
--
OpenAI 标志
GPT-5 nano (minimal) private
400k
专有
8*
--
--
--
--
--
OpenAI 标志
GPT-4 Turbo
128k
专有
8*
--
32
3.01
18.79
--
OpenAI 标志
GPT-4
8.19k
专有
7*
--
--
--
--
--
OpenAI 标志
GPT-4o mini
128k
专有
7*
--
--
--
--
--
OpenAI 标志
GPT-4o mini
128k
专有
7*
--
75
1.10
7.80
--
OpenAI 标志
GPT-3.5 Turbo
4.1k
专有
4*
--
--
--
--
--
OpenAI 标志
GPT-5.4 Pro (xhigh)
1.05M
专有
--
--
--
--
--
--

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 OpenAI 的常见问题

OpenAI 上最智能的模型是 GPT-5.6 Sol (max),Intelligence Index 得分为 59。

按输出速度计算,OpenAI 上最快的模型是 o3-mini,速度为每秒 210.9 个 token。

OpenAI 上首个答案 Token 延迟最低的模型是 GPT-5.4 nano,延迟为 0.58 秒。延迟越低,初始响应越快。

按混合价格计算,OpenAI 上最实惠的模型是 GPT-5 nano (high),每 100 万 token 的价格为 $0.05(缓存命中/输入/输出比例为 7:2:1)。

OpenAI 上各模型价格最多相差 224 倍,从最实惠的 GPT-5 nano (high)(每 100 万 token $0.05)到最昂贵的 GPT-4 Turbo(每 100 万 token $12.00)。

是,OpenAI 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。

是,OpenAI 上全部 61 个模型均支持用于结构化输出的 JSON 模式。

是,OpenAI 上全部 61 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 OpenAI 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。