OpenAI has unveiled a series of major updates and price reductions for its AI models, signaling a step forward in the accessibility and efficiency of AI technology. The updates include new embedding models, enhancements to the GPT-4 Turbo and GPT-3.5 Turbo models, and the introduction of a new 4T model set to launch in the coming months.
OpenAI is introducing two new embedding models: text-embedding-3-small and text-embedding-3-large. Embeddings are sequences of numbers representing concepts within content, such as natural language or code, and are crucial for machine learning models to understand relationships between content.
The text-embedding-3-small model offers a substantial upgrade over its predecessor, text-embedding-ada-002, with improved performance and a 5X reduction in price. The text-embedding-3-large model, on the other hand, creates larger embeddings with up to 3072 dimensions and showcases superior performance.
OpenAI is set to launch a new GPT-3.5 Turbo model, gpt-3.5-turbo-0125, next week. This update comes with a 50% reduction in input costs and a 25% cut in output prices, making it more affordable for customers. The new model will also feature various improvements, including higher accuracy and a fix for a text encoding issue affecting non-English language function calls.
The updated GPT-4 Turbo preview model, gpt-4-0125-preview, aims to complete tasks like code generation more thoroughly and reduce instances of “laziness” where the model doesn’t complete a task. This model also includes a fix for the bug impacting non-English UTF-8 generations. OpenAI plans to launch GPT-4 Turbo with vision in general availability soon.
OpenAI is also releasing text-moderation-007, its most robust moderation model to date. Additionally, new platform improvements will give developers more visibility into their usage and control over API keys, including the ability to assign permissions to API keys and track usage on an API key level.
