chatgpt embeddings price的进一步展开说明

Introduction

OpenAI made several significant updates to its language models, including improved steerability, enhanced function calling capabilities, extended context windows, and lower prices. These updates have been well-received, especially the launch of ChatGPT, a model trained to provide detailed responses based on instructions. In this blog post, we will delve into the details of these updates and their implications.

Improved Function Calling in the Chat Completions API

OpenAI has introduced new function calling capabilities in the Chat Completions API. Developers can now describe functions to the models and receive JSON output containing arguments for calling those functions. This enables chatbots to interact with external tools and APIs to generate structured data output. The models have been fine-tuned to detect when a function needs to be called and can respond with JSON adhering to the function signature. This allows developers to create chatbots that can answer questions, convert natural language into API calls or database queries, and extract structured data from text. The Chat Completions API also supports calling specific functions, and developers can refer to the developer documentation for adding evals to improve function calling.

Function Calling Example

To illustrate the function calling capabilities, consider the following example:

Step 1: OpenAI API
The model is called with functions and the user’s input, as shown in the sample request code.

Step 2: Third Party API
The model response is used to call the developer’s API, as demonstrated in the provided request.

Step 3: OpenAI API
The response from the third-party API is sent back to the model for summarization, as depicted in the sample request code.

The complete response provides the desired output, in this case, the weather in Boston.

Bigger Context Window

The latest updates to GPT⑷⑶2k-0613 and GPT⑶.5-turbo⑴6k have increased the models’ ability to handle larger texts. This enhancement is particularly useful for applications that require processing substantial amounts of text data. GPT⑶.5-turbo⑴6k can now handle approximately 20 pages of text in a single request, four times more than the previous model. This improvement significantly enhances the model’s capability to deal with larger text data, allowing users to upload up to 20 pages of text into ChatGPT at once.

New Models

OpenAI has introduced updated versions of their GPT⑷ and GPT⑶.5 Turbo models. GPT⑷-0613 includes improvements in function-calling capabilities. Similarly, GPT⑷⑶2k-0613 retains the same function-calling improvements as GPT⑷-0613 but also has an extended context length, enabling better comprehension of larger texts.

These updates aim to provide more people with the opportunity to try GPT⑷. OpenAI plans to invite many more users from the waitlist in the coming weeks, with the goal of removing the waitlist entirely for this model. Additionally, GPT⑶.5-Turbo-0613 now includes function calling and improved steerability with the system message.

Furthermore, GPT⑶.5-Turbo⑴6k offers four times the context length of GPT⑶.5-Turbo, providing support for approximately 20 pages of text in a single request. The pricing for GPT⑶.5-Turbo⑴6k is $0.003 per 1K input tokens and $0.004 per 1K output tokens.

OpenAI will upgrade and deprecate the earlier versions of GPT⑷ and GPT⑶.5 Turbo that were announced in March. Applications using the stable model names will be automatically upgraded to the new models on June 27th. Developers can continue to use the older models until September 13th by specifying the appropriate model names in their API requests. OpenAI encourages developers to provide feedback during this transition period.

Lower Pricing

OpenAI has recently announced a significant price reduction for its embeddings model. The costs have been reduced by 75%, resulting in a price of $0.0001 per 1K tokens. This reduction reflects OpenAI’s ongoing efforts to make their systems more efficient and cost-effective for developers.

The price reduction also applies to GPT⑶.5 Turbo, the popular chat model powering ChatGPT. The cost of input tokens for this model has been reduced by 25%, with developers now able to use it for $0.0015 per 1K input tokens and $0.002 per 1K output tokens. The 16k context version, GPT⑶.5-Turbo⑴6k, is priced at $0.003 per 1K input tokens and $0.004 per 1K output tokens.

OpenAI values feedback from developers and relies on their suggestions for the continuous evolution of the platform. These updates provide increased value and a broader range of application opportunities for developers using OpenAI’s models. With lower pricing and other new features, OpenAI remains at the forefront of natural language processing technology.

GPT⑷ Model vs Chat Model (GPT⑶.5 Turbo)

GPT⑶.5-Turbo offers a budget-friendly option with sufficient performance for general applications and lower resource requirements. This makes it ideal for those on a limited budget or working in resource-constrained environments. However, its context window is limited, and its function calling capability is less powerful compared to GPT⑷, making it less suitable for advanced applications.

On the other hand, GPT⑷ provides enhanced function calling capabilities and a larger context window, which is beneficial for complex applications that require retaining extensive past information. However, it comes at a higher price point and requires greater computational resources. Choosing the right model depends on the specific needs and budget of the project.

Model Deprecation

OpenAI has initiated a process of upgrading and deprecating the initial versions of GPT⑷ and GPT⑶.5 Turbo, which were announced in March. The stable model names, including gpt⑶.5-turbo, gpt⑷, and gpt⑷⑶2k, will be automatically upgraded to the new models on June 27th. To compare model performance between versions, OpenAI’s Evals library supports public and private evaluations to demonstrate how model changes may impact specific use cases.

For developers who require more time to transition to the new models, the older models (gpt⑶.5-turbo-0301, gpt⑷-0314, or gpt⑷⑶2k-0314) can still be used until September 13th by specifying the appropriate model names in API requests. After September 13th, requests using these model names will be unsuccessful. OpenAI encourages developers to stay updated by visiting the model deprecation page and welcomes feedback to ensure a smooth transition.

Conclusion

OpenAI’s latest updates to their language models offer powerful capabilities for developers. The reduced pricing for both the embeddings model and GPT⑶.5 Turbo provides opportunities for more experimentation with web apps while minimizing expenses. The added function calling capabilities and larger context window push the boundaries of natural language processing and put cutting-edge technology in the hands of developers worldwide.

In addition, the ChatGPT plugin is an AI-powered tool that assists in handling customer service, marketing, and other business tasks. This article has introduced various creative use cases to bring more value to your business, and OpenAI remains committed to continually improving and refining their models based on developer feedback.

chatgpt embeddings price的常见问答Q&A