OpenAI Unveils Upgraded AI Models, Lowers Prices and Plans for Creator Monetization
Summary:
OpenAI has introduced updated AI models, including a superior GPT-4 Turbo for improved task completion and a fresh GPT-3.5 Turbo model with drastically reduced prices. Developers now have improved ways to manage API keys and usage. Smaller AI models known as 'embeddings' have also been unveiled, facilitating computers to comprehend and process written text more efficiently. The company also revealed its intention to allow GPT creators to monetize their AI systems.
OpenAI has unveiled its updated GPT-4 Turbo model, featuring improved task completion, and a significant decrease in the cost of GPT-3.5 Turbo API. The artificial intelligence research company has also presented fresh methods for developers to manage API keys and assess API usage. The company's blog reveals that the GPT-4 Turbo is more proficient in tasks like code generation than the former model and will help mitigate instances of the model not completing tasks, a problem colloquially referred to as "laziness".
Also revealed is the introduction of a new GPT-3.5 Turbo model, named gpt-3.5-turbo-0125, with drastically reduced price tags to make scaling more achievable for customers. Input costs for this model have been slashed by half to $0.0005 per 1000 tokens, and the output cost lessened by a quarter to $0.0015 per 1000 tokens.
Last December, ChatGPT users expressed frustration about the chatbot often refusing tasks, citing the lag in GPT-4 updates as the problem. While GPT-4 still exhibits some "laziness", the recently launched GPT-4 Turbo should alleviate this issue significantly as it was trained on data up until April 2023.
OpenAI also unveiled smaller AI models, termed embeddings, that enable computers to process and understand written text more efficiently by converting it into a processable format, acting a lot like a translator capable of interpreting human language.
A type of AI that uses these embeddings, referred to as Retrieval-augmented generation, has also been introduced. Rather than generating responses from scratch, it recalls information much like referring to a reference book, ensuring more accurate and pertinent responses. Two new models of these embeddings are now available. "text-embedding-3-small" and a more potent counterpart, "text-embedding-3-large", refer to their processing capacities.
Compared to GPT-4, other AI models like Google's Gemini excel in advanced mathematical and specialized coding tasks. However, there is a debate suggesting that if GPT-4 Turbo were used against Gemini, the outcome might vary.
OpenAI also aims to reward GPT creators by monetizing their customized AI systems. The earnings will be based on user interaction with their GPTs. Nevertheless, the GPT store launch will initially be aimed at users with paid ChatGPT plans.
Published At
1/26/2024 2:08:40 PM
Disclaimer: Algoine does not endorse any content or product on this page. Readers should conduct their own research before taking any actions related to the asset, company, or any information in this article and assume full responsibility for their decisions. This article should not be considered as investment advice. Our news is prepared with AI support.
Do you suspect this content may be misleading, incomplete, or inappropriate in any way, requiring modification or removal?
We appreciate your report.