In a groundbreaking advancement, OpenAI has introduced GPT-4o mini, the newest cost-efficient model designed to democratize access to artificial intelligence. With enhancements in affordability and performance, GPT-4o mini is positioned to broaden AI application horizons significantly. This model is not only cost-effective, priced at 15 cents per million input tokens and 60 cents per million output tokens, but it also surpasses its predecessors like GPT-3.5 Turbo by over 60% in cost efficiency.
Remarkably, GPT-4o mini excels across a spectrum of metrics, including reasoning tasks where it scored 82% on the MMLU benchmark, to outperforming other models in math reasoning with an 87% score on MGSM. The model's adeptness extends to coding performance as evidenced by its 87.2% rating on HumanEval.
The versatility of GPT-4o mini is seen in its robust support for text and multimodal interactions, covering up to 128K tokens per context window. Future updates promise even broader multimedia support.
With safety and performance underscored by collaborative evaluations, the GPT-4o mini redefines AI reliability and affordability. For an in-depth exploration of GPT-4o mini's potential, visit OpenAI’s detailed blog post: Explore GPT-4o Mini.
Source: https://openai.com/blog/gpt-4o-mini-advancing-cost-efficient-intelligence
Top comments (0)