■Key Features

1. Fast and Advanced Language Understanding and Generation
GPT-4o mini outperforms GPT-3.5 Turbo in text intelligence (scoring 82% on MMLU compared to 69.8% for GPT-3.5 Turbo) and multimodal reasoning, demonstrating high performance.
Additionally, Japanese processing speed has improved, enabling more natural conversations with instant responses.


2. Enhanced Context Understanding
It supports a 128K token input context and can generate outputs of up to 16K tokens. Trained on information up to October 2023, it can handle relatively recent information.


3. More Accurate and Detailed Responses
GPT-4o mini is about 40% more likely to provide fact-based answers compared to GPT-3.5 Turbo, understanding more complex problems and providing higher quality responses.