OpenAI is launching GPT-4o Mini, a lighter and cheaper model for developers. It aims to make AI more accessible by offering a more affordable alternative to the expensive full-sized models, while being more capable than GPT-3.5. ChatGPT users on Free, Plus and Team plans can use GPT-4o Mini for free instead of GPT-3.5 Turbo, with Enterprise users getting access week later. GPT-3.5 will eventually be retired from the API.
The new model supports text and vision in the API and will soon handle multimodal inputs and outputs like video and audio. GPT-4o Mini scored 82% on the MMLU benchmark, compared to GPT-3.5’s 70% and GPT-4o’s 88.7%. Despite concerns about benchmark test reliability, GPT-4o Mini offers a cost-effective tool for developers to build AI applications.
OpenAI’s decision to prioritize smaller models like GPT-4o Mini reflects a growing demand from developers for affordable AI solutions. The model has already been tested by startups like Ramp and Superhuman, demonstrating its practical applications in expense data extraction and email auto-suggestions.
Source: The Verge