Monday, February 26, 2024

OpenAI Unveils Major API Updates: Introduces Steerable GPT-3.5-Turbo and GPT-4

Share

OpenAI recently made several API updates, introducing steerable GPT-3.5-Turbo and GPT-4 models, function calling support in the Chat Completions API, and lower pricing. These updates bring significant improvements to the capabilities and performance of the models.

One notable addition is the function calling capability in the Chat Completions API, enabling dynamic and interactive conversations. Developers can now utilize this feature to make the models detect and call functions, receiving JSON responses that conform to the specified function requirements. This enhancement allows for more versatile interactions with the AI models.

OpenAI has also taken steps to make its models more accessible and affordable. They have implemented a 75% cost reduction on their state-of-the-art embeddings model, making it more cost-effective for developers. Additionally, there is a 25% cost reduction on input tokens specifically for GPT-3.5-Turbo. These price reductions aim to enable users to achieve more within their budget while leveraging the capabilities of the models.

These API updates reflect OpenAI’s commitment to enhancing their offerings based on developer feedback and industry needs. The improved functionality, increased context length, and lower pricing make it easier for developers to utilize the models effectively and create innovative applications that leverage the power of AI.

Applications Powered By OpenAI

OpenAI’s recent advancements in GPT models not only benefit developers but also have a significant impact on various applications. Several companies have already integrated OpenAI’s generative AI into their products and services, harnessing the power of these models. Microsoft has partnered with OpenAI to enhance Bing and Office with generative AI, while Snapchat has introduced My AI, a chatbot utilizing OpenAI GPT models that can now send and interpret image snaps. Salesforce has released Einstein GPT, the first generative AI CRM product powered by OpenAI’s advanced models. Additionally, Morgan Stanley, HubSpot, GitHub Copilot, Stripe, GetResponse, and Instacart have all leveraged OpenAI technology to enhance their respective offerings.

Enhancements To GPT-3.5 Turbo And GPT-4

OpenAI has made significant improvements to their GPT-3.5 Turbo and GPT-4 models. These enhancements include a new function calling capability in the Chat Completions API, improved steerability, extended context for GPT-3.5 Turbo, and lower pricing. In particular, OpenAI introduced a new 16k context version of GPT-3.5-Turbo, which extends the context length compared to the standard 4k version. This expanded context capacity empowers the models to comprehend and generate responses for much larger bodies of text, providing a more comprehensive understanding of the input.

Function Calling Capability

The function calling capability allows developers to describe functions to the models and receive a JSON object containing arguments intelligently generated by AI. This enhancement facilitates better integration of GPT’s capabilities with external tools and APIs, enabling structured data retrieval from the model. The new API parameters open up possibilities such as creating chatbots that utilize external tools, converting natural language queries into function calls or API/database queries, and extracting structured data from text. Developers can seamlessly integrate GPT models with other APIs and external tools, making applications more user-friendly and efficient.

Model Improvements

OpenAI has also improved steerability and extended the context in GPT-4 and GPT-3.5 Turbo models. Increased steerability allows developers to design AI applications that align more closely with specific organizational or task requirements, resulting in targeted business reports or context-aware responses in customer service chatbots. The extended context capacity of GPT-3.5 Turbo-16k enables the model to comprehend and generate responses for much larger bodies of text, supporting up to 20 pages of text in a single request. This extended context is beneficial for tasks involving lengthy documents or project management applications, where the AI can process and understand large amounts of text, leading to more efficient information extraction and insightful project analytics.

OpenAI is also focused on improving affordability. They have reduced the pricing of the text-embedding-ada-002 model by 75% and introduced a 25% cost reduction on input tokens for the GPT-3.5 Turbo model. These price reductions, coupled with the enhanced functionality, make it easier for developers to utilize and experiment with these models in their applications.

OpenAI continues to emphasize the development of its GPT models based on valuable feedback from developers. By providing these latest enhancements, OpenAI empowers developers to create innovative and improved AI applications that address the complexity and specificity of tasks commonly encountered in workplace environments. The updated API and GPT models offer developers enhanced capabilities to build AI applications that can effectively handle a wide range of workplace challenges.

Read more

Recommended For You