New Delhi: OpenAI has unveiled an updated version of its AI models and pricing, aiming to enhance user experience and streamline operations. The GPT-3.5 Turbo model now comes with a 50% reduction in per-token pricing for input and 25% for output, making it more affordable for developers involved in text-intensive tasks like document or book analysis. The company’s pricing strategy aligns with the evolving landscape of open-source models.
The GPT-3.5 Turbo has been upgraded to version 0125, incorporating unspecified improvements from its previous version, 0613. Although details about the enhancements are not explicitly provided, OpenAI’s commitment to refining its models is evident.
Additionally, a preview model version 0125 has been introduced for GPT-4 Turbo, with a focus on addressing “laziness” issues in tasks such as code generation. The objective is to improve performance on technical tasks compared to earlier versions. While the GPT-4 with vision capabilities is currently in preview, a broader launch is anticipated in the coming months.
The upgraded text embedding models offer more effective semantic representations of language, catering to the needs of researchers and engineers. While these adjustments may not be groundbreaking, they highlight OpenAI’s ongoing efforts to enhance its AI offerings incrementally.
