OpenAI, has introduced new models to the public, including the updated GPT-4 Turbo preview model, and has taken steps to meet user demand by reducing the price of GPT-3.5 Turbo application programming interface (API) access. The company also mentioned that it provided new ways for developers to manage their API keys and understand API usage.
Noteworthy Steps from the OpenAI Team
OpenAI team in a blog post explained that the updated GPT-4 Turbo completes tasks such as code generation more comprehensively than the previous preview model and aims to reduce instances of the model failing to complete a task.
OpenAI also introduced a new GPT-3.5 Turbo model called gpt-3.5-turbo-0125 and announced it would reduce GPT-3.5 Turbo prices for the third time in the past year to help customers scale. The new model’s prices were reduced by 50% to $0.0005 per thousand tokens, and output prices were reduced by 25% to $0.0015 per thousand tokens.
ChatGPT users had complained in December 2023 that the chatbot frequently refused tasks, which was attributed to GPT-4 not being updated. While GPT-4 users using data available before September 2021 may still experience the same issues, GPT-4 Turbo was trained on new information up to April 2023.
The OpenAI team also introduced smaller artificial intelligence models known as embeddings. OpenAI defines embeddings as sequences of numbers that represent concepts in content such as language or code.
Artificial Intelligence Models
Embeddings, are a type of artificial intelligence tool that helps computers understand and use written text more effectively. They do this by converting words and sentences into a format that computers can process. Think of it as a translator that turns human language into a special code that computers can understand and work with.
Retrieval-augmented is a type of artificial intelligence that provides more accurate and relevant answers instead of generating responses from scratch. It’s like having an AI that quickly looks up a reference book and tells you what it found instead of guessing an answer.
Currently, there are two new models available that use these embeddings: text-embedding-3-small and a more powerful version called text-embedding-3-large. The small and large designations indicate the capacity of these models. The larger model works like a more comprehensive translator and can understand and convert text more clearly than the smaller one. These models can now be used in applications that need to efficiently retrieve and use information.
