OpenAI, a leading company in the generative AI space, has recently unveiled upgraded text-generating models and revised Pricing plans to stay competitive. The company's latest offerings include GPT-3.5-turbo and GPT-4, with the latter introducing a new capability called function calling. These advancements open up new possibilities for developers to leverage AI in programming and natural language processing tasks.

Function calling allows developers to describe programming functions to GPT-3.5-turbo and GPT-4, enabling the models to generate code that executes the specified functions. This functionality is particularly useful for developing chatbots that can utilize external tools, convert natural language into database queries, and extract structured data from text. OpenAI has fine-tuned these models to recognize when a function needs to be called and respond with JSON that adheres to the function signature, providing reliable structured data output.

In addition to function calling, OpenAI has introduced an enhanced version of GPT-3.5-turbo with an expanded context window. The context window refers to the amount of text the model considers before generating subsequent text. By increasing the context length fourfold to 16,000 tokens, GPT-3.5-turbo with the expanded context window improves its ability to retain and understand the context of conversations, minimizing the risk of going off-topic.

The pricing structure for OpenAI's models has also undergone changes. The new GPT-3.5-turbo with the expanded context window is available at $0.003 per 1,000 input tokens and $0.004 per 1,000 output tokens, offering improved capabilities at a slightly higher cost. This version can process approximately 20 pages of text in a single run, but it falls short of the processing capacity of AI startup Anthropic's flagship model.

On a positive note, OpenAI has reduced the pricing for the original GPT-3.5-turbo model by 25%. Developers can now access this model at $0.0015 per 1,000 input tokens and $0.002 per 1,000 output tokens, making it more cost-effective. With this revised pricing, developers can obtain approximately 700 pages of processed text per dollar spent.

OpenAI has also reduced the pricing of its popular text embedding model, text-embedding-ada-002, by 75%. Text embeddings measure the relatedness of text strings and are widely used for search and recommendation systems. The updated pricing for text-embedding-ada-002 is now $0.0001 per 1,000 tokens, making it more accessible to developers looking to leverage text-related AI tasks.

These changes in pricing and new generative text features aim to enhance OpenAI's offerings and provide developers with more advanced tools at competitive prices. By reducing pricing for certain models and introducing additional functionalities, OpenAI continues to empower developers and researchers in their pursuit of leveraging AI for various applications.

Reference:

URL: https://techcrunch.com/2023/06/13/openai-intros-new-generative-text-features-while-reducing-pricing/