The OpenAI API is powered by a diverse set of models with different capabilities and price points. You can also make customizations to our models for your specific use case with fine-tuning.
OpenAI API 由一系列具有不同功能和价位的模型提供支持。您还可以根据自己的具体使用情况对我们的模型进行定制和微调。
Model | Description |
---|---|
GPT-4 Turbo and GPT-4 GPT-4 Turbo 和 GPT-4 | A set of models that improve on GPT-3.5 and can understand as well as generate natural language or code 一套在 GPT-3.5 基础上改进的模型,可理解并生成自然语言或代码 |
GPT-3.5 Turbo GPT-3.5 涡轮增压 | A set of models that improve on GPT-3.5 and can understand as well as generate natural language or code 一套在 GPT-3.5 基础上改进的模型,可理解并生成自然语言或代码 |
DALL·E | A model that can generate and edit images given a natural language prompt 能根据自然语言提示生成和编辑图像的模型 |
TTS | A set of models that can convert text into natural sounding spoken audio 一套可将文本转换成自然发音口语音频的模型 |
Whisper | A model that can convert audio into text 可将音频转换成文本的模型 |
Embeddings | A set of models that can convert text into a numerical form 一套可将文本转换为数字形式的模型 |
Moderation | A fine-tuned model that can detect whether text may be sensitive or unsafe 可检测文本是否敏感或不安全的微调模型 |
GPT base | A set of models without instruction following that can understand as well as generate natural language or code 一套不遵循指令的模型,可理解并生成自然语言或代码 |
Deprecated | A full list of models that have been deprecated along with the suggested replacement 已弃用模型的完整列表以及建议的替代方案 |
We have also published open source models including Point-E, Whisper, Jukebox, and CLIP.
我们还发布了开源模型,包括 Point-E、Whisper、Jukebox 和 CLIP。
gpt-4-turbo
, gpt-4
, and gpt-3.5-turbo
point to their respective latest model version. You can verify this by looking at the response object after sending a request. The response will include the specific model version used (e.g. gpt-3.5-turbo-0613
).
gpt-4-turbo
、 和 指向各自的最新型号版本。您可以通过查看发送请求后的响应对象来验证这一点。响应将包括所使用的特定型号版本(如 )。 gpt-4
gpt-3.5-turbo
gpt-3.5-turbo-0613
We also offer pinned model versions that developers can continue using for at least three months after an updated model has been introduced. With the new cadence of model updates, we are also giving people the ability to contribute evals to help us improve the model for different use cases. If you are interested, check out the OpenAI Evals repository.
我们还提供固定模型版本,开发人员可以在更新模型推出后至少三个月内继续使用。随着模型更新的新节奏,我们还让人们能够贡献 Evals,以帮助我们针对不同的用例改进模型。如果您对此感兴趣,请访问 OpenAI Evals 信息库。
Learn more about model deprecation on our deprecation page.
有关模型弃用的更多信息,请访问我们的弃用页面。
GPT-4 is a large multimodal model (accepting text or image inputs and outputting text) that can solve difficult problems with greater accuracy than any of our previous models, thanks to its broader general knowledge and advanced reasoning capabilities. GPT-4 is available in the OpenAI API to paying customers. Like gpt-3.5-turbo
, GPT-4 is optimized for chat but works well for traditional completions tasks using the Chat Completions API. Learn how to use GPT-4 in our text generation guide.
GPT-4 是一个大型多模态模型(接受文本或图像输入并输出文本),凭借其更广泛的常识和高级推理能力,它比我们以前的任何模型都能更准确地解决难题。GPT-4 可通过 OpenAI API 向付费用户提供。与 gpt-3.5-turbo
一样,GPT-4 针对聊天进行了优化,但使用聊天完成 API 也能很好地完成传统的完成任务。在我们的文本生成指南中了解如何使用 GPT-4。
Model | Description | Context window 背景窗口 | Training data 培训数据 |
---|---|---|---|
gpt-4-turbo | New GPT-4 Turbo with Vision新型 GPT-4 涡轮增压发动机,带远景功能 The latest GPT-4 Turbo model with vision capabilities. Vision requests can now use JSON mode and function calling. Currently points to gpt-4-turbo-2024-04-09 .具有视觉功能的最新 GPT-4 Turbo 型号。视觉请求现在可以使用 JSON 模式和函数调用。目前指向 gpt-4-turbo-2024-04-09 。 | 128,000 tokens 128,000 个代币 | Up to Dec 2023 截至 2023 年 12 月 |
gpt-4-turbo-2024-04-09 | GPT-4 Turbo with Vision model. Vision requests can now use JSON mode and function calling. gpt-4-turbo currently points to this version.带有 Vision 模型的 GPT-4 Turbo。Vision 请求现在可以使用 JSON 模式和函数调用。 gpt-4-turbo 目前指向该版本。 | 128,000 tokens 128,000 个代币 | Up to Dec 2023 截至 2023 年 12 月 |
gpt-4-turbo-preview | GPT-4 Turbo preview model. Currently points to gpt-4-0125-preview .GPT-4 Turbo 预览模型。目前指向 gpt-4-0125-preview 。 | 128,000 tokens 128,000 个代币 | Up to Dec 2023 截至 2023 年 12 月 |
gpt-4-0125-preview | GPT-4 Turbo preview model intended to reduce cases of “laziness” where the model doesn’t complete a task. Returns a maximum of 4,096 output tokens. Learn more. GPT-4 Turbo 预览模型旨在减少模型未完成任务的 "懒惰 "情况。最多可返回 4,096 个输出标记。了解更多 | 128,000 tokens 128,000 个代币 | Up to Dec 2023 截至 2023 年 12 月 |
gpt-4-1106-preview | GPT-4 Turbo preview model featuring improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more. Returns a maximum of 4,096 output tokens. This is a preview model. Learn more. GPT-4 Turbo 预览模型具有改进的指令跟踪、JSON 模式、可重现输出、并行函数调用等功能。最多可返回 4,096 个输出标记。此为预览机型。了解更多信息。 | 128,000 tokens 128,000 个代币 | Up to Apr 2023 至 2023 年 4 月 |
gpt-4-vision-preview | GPT-4 model with the ability to understand images, in addition to all other GPT-4 Turbo capabilities. This is a preview model, we recommend developers to now use gpt-4-turbo which includes vision capabilities. Currently points to gpt-4-1106-vision-preview .除了 GPT-4 Turbo 的所有其他功能外,GPT-4 模型还具有理解图像的功能。这是一个预览模型,我们建议开发人员现在使用包含视觉功能的 gpt-4-turbo 。目前指向 gpt-4-1106-vision-preview 。 | 128,000 tokens 128,000 个代币 | Up to Apr 2023 至 2023 年 4 月 |
gpt-4-1106-vision-preview | GPT-4 model with the ability to understand images, in addition to all other GPT-4 Turbo capabilities. This is a preview model, we recommend developers to now use gpt-4-turbo which includes vision capabilities. Returns a maximum of 4,096 output tokens. Learn more.除了 GPT-4 Turbo 的所有其他功能外,GPT-4 模型还具有理解图像的功能。这是一个预览模型,我们建议开发人员现在使用包含视觉功能的 gpt-4-turbo 。最多可返回 4,096 个输出标记。了解更多信息。 | 128,000 tokens 128,000 个代币 | Up to Apr 2023 至 2023 年 4 月 |
gpt-4 | Currently points to gpt-4-0613 . See continuous model upgrades.目前指向 gpt-4-0613 。请参见持续的型号升级。 | 8,192 tokens | Up to Sep 2021 至 2021 年 9 月 |
gpt-4-0613 | Snapshot of gpt-4 from June 13th 2023 with improved function calling support.2023 年 6 月 13 日的 gpt-4 快照,改进了函数调用支持。 | 8,192 tokens | Up to Sep 2021 至 2021 年 9 月 |
gpt-4-32k | Currently points to gpt-4-32k-0613 . See continuous model upgrades. This model was never rolled out widely in favor of GPT-4 Turbo.目前指向 gpt-4-32k-0613 。请参见持续的型号升级。该型号从未广泛推广,而是采用了 GPT-4 Turbo。 | 32,768 tokens 32 768 个代币 | Up to Sep 2021 至 2021 年 9 月 |
gpt-4-32k-0613 | Snapshot of gpt-4-32k from June 13th 2023 with improved function calling support. This model was never rolled out widely in favor of GPT-4 Turbo.2023 年 6 月 13 日的 gpt-4-32k 快照,改进了函数调用支持。该型号从未广泛推广,而是采用了 GPT-4 Turbo。 | 32,768 tokens 32 768 个代币 | Up to Sep 2021 至 2021 年 9 月 |
For many basic tasks, the difference between GPT-4 and GPT-3.5 models is not significant. However, in more complex reasoning situations, GPT-4 is much more capable than any of our previous models.
在许多基本任务中,GPT-4 和 GPT-3.5 模型之间的差异并不大。然而,在更复杂的推理情况下,GPT-4 比我们之前的任何模型都更有能力。
GPT-4 outperforms both previous large language models and as of 2023, most state-of-the-art systems (which often have benchmark-specific training or hand-engineering). On the MMLU benchmark, an English-language suite of multiple-choice questions covering 57 subjects, GPT-4 not only outperforms existing models by a considerable margin in English, but also demonstrates strong performance in other languages.
GPT-4 的性能不仅优于以前的大型语言模型,而且截至 2023 年,还优于大多数最先进的系统(这些系统通常有特定的基准训练或手工工程)。在涵盖 57 个科目的英语多选题套件 MMLU 基准测试中,GPT-4 不仅在英语方面远远超过了现有模型,而且在其他语言方面也表现出色。
GPT-3.5 Turbo models can understand and generate natural language or code and have been optimized for chat using the Chat Completions API but work well for non-chat tasks as well.
GPT-3.5 Turbo 模型可以理解和生成自然语言或代码,并使用聊天完成 API 对聊天进行了优化,但也能很好地用于非聊天任务。
Model | Description | Context window 背景窗口 | Training data 培训数据 |
---|---|---|---|
gpt-3.5-turbo-0125 | New Updated GPT 3.5 Turbo最新更新的 GPT 3.5 Turbo The latest GPT-3.5 Turbo model with higher accuracy at responding in requested formats and a fix for a bug which caused a text encoding issue for non-English language function calls. Returns a maximum of 4,096 output tokens. Learn more. 最新的 GPT-3.5 Turbo 模型能更准确地响应所要求的格式,并修复了一个导致非英语语言函数调用文本编码问题的错误。最多可返回 4,096 个输出标记。了解更多信息。 | 16,385 tokens 16,385 枚代币 | Up to Sep 2021 至 2021 年 9 月 |
gpt-3.5-turbo | Currently points to gpt-3.5-turbo-0125 .目前指向 gpt-3.5-turbo-0125 。 | 16,385 tokens 16,385 枚代币 | Up to Sep 2021 至 2021 年 9 月 |
gpt-3.5-turbo-1106 | GPT-3.5 Turbo model with improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more. Returns a maximum of 4,096 output tokens. Learn more. GPT-3.5 Turbo 模型改进了指令跟踪、JSON 模式、可重现输出、并行函数调用等功能。最多可返回 4,096 个输出标记。了解更多信息。 | 16,385 tokens 16,385 枚代币 | Up to Sep 2021 至 2021 年 9 月 |
gpt-3.5-turbo-instruct | Similar capabilities as GPT-3 era models. Compatible with legacy Completions endpoint and not Chat Completions. 与 GPT-3 时代的型号功能相似。与传统的 Completions 终端兼容,不兼容 Chat Completions。 | 4,096 tokens | Up to Sep 2021 至 2021 年 9 月 |
gpt-3.5-turbo-16k | Legacy Currently points to gpt-3.5-turbo-16k-0613 .遗产目前指向 gpt-3.5-turbo-16k-0613 。 | 16,385 tokens 16,385 枚代币 | Up to Sep 2021 至 2021 年 9 月 |
gpt-3.5-turbo-0613 | Legacy Snapshot of gpt-3.5-turbo from June 13th 2023. Will be deprecated on June 13, 2024.2023 年 6 月 13 日的 gpt-3.5-turbo 传统快照。将于 2024 年 6 月 13 日废弃。 | 4,096 tokens | Up to Sep 2021 至 2021 年 9 月 |
gpt-3.5-turbo-16k-0613 | Legacy Snapshot of gpt-3.5-16k-turbo from June 13th 2023. Will be deprecated on June 13, 2024.2023 年 6 月 13 日的 gpt-3.5-16k-turbo 传统快照。将于 2024 年 6 月 13 日废弃。 | 16,385 tokens 16,385 枚代币 | Up to Sep 2021 至 2021 年 9 月 |
DALL·E is a AI system that can create realistic images and art from a description in natural language. DALL·E 3 currently supports the ability, given a prompt, to create a new image with a specific size. DALL·E 2 also support the ability to edit an existing image, or create variations of a user provided image.
DALL-E 是一个人工智能系统,可以根据自然语言的描述创建逼真的图像和艺术品。DALL-E 3 目前支持根据提示创建具有特定尺寸的新图像。DALL-E 2 还支持编辑现有图像,或在用户提供的图像基础上进行修改。
DALL·E 3 is available through our Images API along with DALL·E 2. You can try DALL·E 3 through ChatGPT Plus.
DALL-E 3 可通过我们的图像 API 与 DALL-E 2 一同使用。您可以通过 ChatGPT Plus 试用 DALL-E 3。
Model | Description |
---|---|
dall-e-3 | New DALL·E 3 新《达利尔 3The latest DALL·E model released in Nov 2023. Learn more. 2023 年 11 月发布的最新 DALL-E 型号。了解更多 |
dall-e-2 | The previous DALL·E model released in Nov 2022. The 2nd iteration of DALL·E with more realistic, accurate, and 4x greater resolution images than the original model. 上一个 DALL-E 模型于 2022 年 11 月发布。第二次迭代的 DALL-E 模型更加逼真、精确,图像分辨率是原模型的 4 倍。 |
TTS is an AI model that converts text to natural sounding spoken text. We offer two different model variates, tts-1
is optimized for real time text to speech use cases and tts-1-hd
is optimized for quality. These models can be used with the Speech endpoint in the Audio API.
TTS 是一种人工智能模型,可将文本转换为自然发音的口语文本。我们提供两种不同的模型变体: tts-1
针对实时文本到语音使用案例进行了优化,而 tts-1-hd
则针对质量进行了优化。这些模型可与音频 API 中的语音端点一起使用。
Model | Description |
---|---|
tts-1 | New Text-to-speech 1 新文本转语音 1The latest text to speech model, optimized for speed. 最新的文本到语音模式,优化了速度。 |
tts-1-hd | New Text-to-speech 1 HD 新文本转语音 1 HDThe latest text to speech model, optimized for quality. 最新的文本到语音模式,质量经过优化。 |
Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multi-task model that can perform multilingual speech recognition as well as speech translation and language identification. The Whisper v2-large model is currently available through our API with the whisper-1
model name.
Whisper 是一种通用语音识别模型。它是在一个大型的各种音频数据集上训练出来的,也是一个多任务模型,可以进行多语言语音识别以及语音翻译和语言识别。Whisper v2-large 模型目前可通过我们的应用程序接口获取,模型名称为 whisper-1
。
Currently, there is no difference between the open source version of Whisper and the version available through our API. However, through our API, we offer an optimized inference process which makes running Whisper through our API much faster than doing it through other means. For more technical details on Whisper, you can read the paper.
目前,Whisper 的开源版本与通过我们的应用程序接口提供的版本没有区别。不过,通过我们的应用程序接口,我们提供了优化的推理过程,这使得通过我们的应用程序接口运行 Whisper 要比通过其他方式快得多。有关 Whisper 的更多技术细节,请阅读论文。
Embeddings are a numerical representation of text that can be used to measure the relatedness between two pieces of text. Embeddings are useful for search, clustering, recommendations, anomaly detection, and classification tasks. You can read more about our latest embedding models in the announcement blog post.
嵌入是文本的一种数字表示法,可用于衡量两段文本之间的相关性。嵌入对搜索、聚类、推荐、异常检测和分类任务非常有用。您可以在公告博文中阅读有关我们最新嵌入模型的更多信息。
Model | Description | Output Dimension 输出尺寸 |
---|---|---|
text-embedding-3-large | New Embedding V3 large 新的嵌入式 V3 大Most capable embedding model for both english and non-english tasks 最适合英语和非英语任务的嵌入模型 | 3,072 |
text-embedding-3-small | New Embedding V3 small 新的嵌入式 V3 小版本Increased performance over 2nd generation ada embedding model 性能超过第二代 ada 嵌入模型 | 1,536 |
text-embedding-ada-002 文本嵌入--ADA-002 | Most capable 2nd generation embedding model, replacing 16 first generation models 功能最强大的第二代嵌入式模型,取代 16 个第一代模型 | 1,536 |
The Moderation models are designed to check whether content complies with OpenAI's usage policies. The models provide classification capabilities that look for content in the following categories: hate, hate/threatening, self-harm, sexual, sexual/minors, violence, and violence/graphic. You can find out more in our moderation guide.
审核模型旨在检查内容是否符合 OpenAI 的使用政策。这些模型提供分类功能,可查找以下类别的内容:仇恨、仇恨/威胁、自残、性、性/未成年人、暴力和暴力/图片。您可以在我们的审核指南中了解更多信息。
Moderation models take in an arbitrary sized input that is automatically broken up into chunks of 4,096 tokens. In cases where the input is more than 32,768 tokens, truncation is used which in a rare condition may omit a small number of tokens from the moderation check.
节制模型接收任意大小的输入,并自动分割成 4,096 个词块。如果输入内容超过 32 768 个词组,则会使用截断方法,在极少数情况下,这种方法可能会在节制检查中忽略少量词组。
The final results from each request to the moderation endpoint shows the maximum value on a per category basis. For example, if one chunk of 4K tokens had a category score of 0.9901 and the other had a score of 0.1901, the results would show 0.9901 in the API response since it is higher.
每次请求审核端点的最终结果都会显示每个类别的最大值。例如,如果 4K 标记中的一个标记块的类别得分是 0.9901,而另一个标记块的类别得分是 0.1901,那么在 API 响应中,结果将显示 0.9901,因为它更高。
Model | Description | Max tokens |
---|---|---|
text-moderation-latest | Currently points to text-moderation-007 .目前指向 text-moderation-007 。 | 32,768 |
text-moderation-stable | Currently points to text-moderation-007 .目前指向 text-moderation-007 。 | 32,768 |
text-moderation-007 | Most capable moderation model across all categories. 所有类别中最有能力的调节模式。 | 32,768 |
GPT base models can understand and generate natural language or code but are not trained with instruction following. These models are made to be replacements for our original GPT-3 base models and use the legacy Completions API. Most customers should use GPT-3.5 or GPT-4.
GPT 基本模型可以理解和生成自然语言或代码,但不进行指令跟踪训练。这些模型可替代我们最初的 GPT-3 基本模型,并使用传统的 Completions API。大多数客户应使用 GPT-3.5 或 GPT-4。
Model | Description | Max tokens | Training data 培训数据 |
---|---|---|---|
babbage-002 | Replacement for the GPT-3 ada and babbage base models.替代 GPT-3 ada 和 babbage 基本型号。 | 16,384 tokens 16,384 枚代币 | Up to Sep 2021 至 2021 年 9 月 |
davinci-002 | Replacement for the GPT-3 curie and davinci base models.替代 GPT-3 curie 和 davinci 基本型号。 | 16,384 tokens 16,384 枚代币 | Up to Sep 2021 至 2021 年 9 月 |
Your data is your data.
你的数据就是你的数据。
As of March 1, 2023, data sent to the OpenAI API will not be used to train or improve OpenAI models (unless you explicitly opt in). One advantage to opting in is that the models may get better at your use case over time.
自 2023 年 3 月 1 日起,发送到 OpenAI API 的数据将不会用于训练或改进 OpenAI 模型(除非您明确选择加入)。选择加入的一个好处是,随着时间的推移,模型在您的使用案例中可能会变得更好。
To help identify abuse, API data may be retained for up to 30 days, after which it will be deleted (unless otherwise required by law). For trusted customers with sensitive applications, zero data retention may be available. With zero data retention, request and response bodies are not persisted to any logging mechanism and exist only in memory in order to serve the request.
为帮助识别滥用行为,API 数据最多可保留 30 天,之后将被删除(除非法律另有规定)。对于具有敏感应用程序的可信客户,可提供零数据保留。在零数据保留的情况下,请求和响应体不会被持久保存到任何日志机制中,仅存在于内存中,以便为请求提供服务。
Note that this data policy does not apply to OpenAI's non-API consumer services like ChatGPT or DALL·E Labs.
请注意,本数据政策不适用于 OpenAI 的非API 消费者服务,如 ChatGPT 或 DALL-E Labs。
Endpoint | Data used for training 用于培训的数据 | Default retention 默认保留 | Eligible for zero retention 有资格获得零保留 |
---|---|---|---|
/v1/chat/completions * | No | 30 days | Yes, except image inputs* 是,图像输入除外* |
/v1/files | No | Until deleted by customer 直至客户删除 | No |
/v1/assistants | No | Until deleted by customer 直至客户删除 | No |
/v1/threads | No | 60 days * | No |
/v1/threads/messages | No | 60 days * | No |
/v1/threads/runs | No | 60 days * | No |
/v1/threads/runs/steps | No | 60 days * | No |
/v1/images/generations | No | 30 days | No |
/v1/images/edits | No | 30 days | No |
/v1/images/variations | No | 30 days | No |
/v1/embeddings | No | 30 days | Yes |
/v1/audio/transcriptions | No | Zero data retention 零数据保留 | - |
/v1/audio/translations | No | Zero data retention 零数据保留 | - |
/v1/audio/speech | No | 30 days | No |
/v1/fine_tuning/jobs | No | Until deleted by customer 直至客户删除 | No |
/v1/moderations | No | Zero data retention 零数据保留 | - |
/v1/completions | No | 30 days | Yes |
* Image inputs via the gpt-4-turbo
model (or previously gpt-4-vision-preview
) are not eligible for zero retention.
* 通过 gpt-4-turbo
模式(或以前的 gpt-4-vision-preview
)输入的图像不符合零保留的条件。
* For the Assistants API, we are still evaluating the default retention period during the Beta. We expect that the default retention period will be stable after the end of the Beta.
* 对于助手 API,我们仍在评估测试版期间的默认保留期。我们预计在测试版结束后,默认保留期将保持稳定。
For details, see our API data usage policies. To learn more about zero retention, get in touch with our sales team.
有关详情,请参阅我们的 API 数据使用政策。如需了解有关零保留的更多信息,请联系我们的销售团队。
Endpoint | Latest models 最新型号 |
---|---|
/v1/assistants | All GPT-4 and GPT-3.5 Turbo models except gpt-3.5-turbo-0301 supported. The retrieval tool requires gpt-4-turbo-preview (and subsequent dated model releases) or gpt-3.5-turbo-1106 (and subsequent versions).支持除 gpt-3.5-turbo-0301 之外的所有 GPT-4 和 GPT-3.5 Turbo 型号。 retrieval 工具要求使用 gpt-4-turbo-preview (及其后续版本)或 gpt-3.5-turbo-1106 (及其后续版本)。 |
/v1/audio/transcriptions | whisper-1 |
/v1/audio/translations | whisper-1 |
/v1/audio/speech | tts-1 , tts-1-hd |
/v1/chat/completions | gpt-4 and dated model releases, gpt-4-turbo-preview and dated model releases, gpt-4-vision-preview , gpt-4-32k and dated model releases, gpt-3.5-turbo and dated model releases, gpt-3.5-turbo-16k and dated model releases, fine-tuned versions of gpt-3.5-turbo gpt-4 和过时的模型发布、 和过时的模型发布、 、 和过时的模型发布、 和过时的模型发布、 和过时的模型发布、微调版的"......"。 gpt-4-turbo-preview gpt-4-vision-preview gpt-4-32k gpt-3.5-turbo gpt-3.5-turbo-16k gpt-3.5-turbo |
/v1/completions (Legacy) | gpt-3.5-turbo-instruct , babbage-002 , davinci-002 |
/v1/embeddings | text-embedding-3-small , text-embedding-3-large , text-embedding-ada-002 |
/v1/fine_tuning/jobs | gpt-3.5-turbo , babbage-002 , davinci-002 |
/v1/moderations | text-moderation-stable , text-moderation-latest |
/v1/images/generations | dall-e-2 , dall-e-3 |
This list excludes all of our deprecated models.
此列表不包括所有已废弃的模型。