On Thursday, OpenAI announced new models and API updates via a blog post. First, the company announced two new embedding models. Then there are the updated versions of GPT-3.5 Turbo and GPT-4.0 Turbo preview. Finally, the company also announced an updated text moderation model.
New GPT-4 Turbo Preview
The new GPT-4 Turbo preview model completes tasks like code generation better than the previous preview model. Further, it also reduces cases of “laziness,” where the model leaves a task incomplete. “The new model also includes the fix for the bug impacting non-English UTF-8 generations.”
Recently, users have criticized the GPT-4 model for its reluctance to complete tasks. However, the company has mentioned in the blog that the update is for the Turbo preview model. Hence, those using GPT-4 could continue experiencing the issues they faced earlier.
However, it is important to mention that GPT-4 Turbo’s database contains more recent information till April 2023. On the other hand, GPT-4’s database is limited to September 2021. Nonetheless, OpenAI has revealed that over 70 percent of users on the non-Turbo model have migrated to the Turbo model.
OpenAI also plans to make GPT-4 Turbo generally available in the coming months. For those catching up, the model supports a 128k context window, handles image prompts, text-to-speech, offers better accuracy, and functionalities like Code Interpreter.
New GPT-3.5 Turbo Coming Next Week
Next week, OpenAI will introduce a new GPT-3.5 Turbo model at a lower price. The new model costs 50 percent less than the older model, implying that input prices will be $0.0005 per 1K tokens. Output prices, on the other hand, have been reduced by 25 percent to $0.0015 per 1K tokens. The model will also feature general improvements, such as higher accuracy.
New Embedding Models
In the blog post, the company explains embedding as a sequence of numbers representing the concepts within content, such as natural language or code. Embeddings make it easier for machine learning models to understand the connection between content and the tasks they need to perform.
The two new embedding models released by OpenAI include text-embedding-3-small and text-embedding-3-large. These provide better performance at lower prices. Last but not least, the free Moderation API now allows developers to identify potentially harmful text.
You can follow Smartprix on Twitter, Facebook, Instagram, and Google News. Visit smartprix.com for the most recent news, reviews, and tech guides.
