OpenAI, a leader in artificial intelligence, has announced some exciting updates. They’ve released new AI models, including an enhanced version of GPT-4, named GPT-4 Turbo. Additionally, they’ve lowered the cost of accessing the GPT-3.5 Turbo application programming interface (API). OpenAI is also making it easier for developers to manage API keys and track how they use the API.
The updated GPT-4 Turbo model is designed to perform tasks, especially code generation, more effectively than its predecessor. It aims to address the issue of ‘laziness’ – when the model doesn’t fully complete a task. This improvement follows user feedback about the chatbot’s performance.
To support its customers, OpenAI has reduced the prices for GPT-3.5 Turbo for the third time this past year. The new prices are significantly lower, with input costs cut by 50% to $0.0005 per thousand tokens and output costs reduced by 25% to $0.0015 per thousand tokens.
OpenAI has also introduced smaller AI models known as embeddings. These sequences of numbers represent concepts in content, like language or code. Embeddings help computers understand and use written text more effectively by converting it into a computer-readable format.
Two new models using embeddings are now available: “text-embedding-3-small” and the more robust “text-embedding-3-large.” These models are designed for applications that need to retrieve and process information efficiently, offering a smarter and more effective way for computers to interact with human language.
OpenAI’s GPT-4 faces competition from other AI models like Google’s Gemini. While Gemini has shown strengths in performing advanced math and specialized coding, comparisons with GPT-4 Turbo are ongoing.
OpenAI also plans to enable GPT creators to monetize their personalized AI systems. Initially, this will be available to U.S. builders, who will be compensated based on user engagement with their AI models. The GPT store, where these systems will be available, will be first rolled out to users on paid ChatGPT plans.
