OpenAI unveils GPT-4o Mini: affordable, high-performance AI model to replace GPT-3.5 Turbo

OpenAI has introduced GPT-4o mini, an affordable and faster AI model set to replace GPT-3.5 Turbo. This new model is part of a broader trend among AI companies to offer cost-effective solutions for developers, as seen with Anthropic's Claude 3 Haiku and Google's Gemini 1.5 Flash. GPT-4o mini is priced competitively at $0.60 per million output tokens and $0.15 per million input tokens, making it 60% cheaper than its predecessor.

Despite its lower cost, GPT-4o mini delivers strong performance, achieving an 82% score on the MMLU benchmark. This is slightly below the 89% scored by the more expensive GPT-4o, but still higher than the 79% and 75% scored by Google Gemini 1.5 Flash and Claude 3 Haiku, respectively. The model features a context window of 128,000 tokens and a knowledge cutoff of October 2023, making it a robust option for various applications.

Developers can access GPT-4o Mini through APIs, while consumers can use it via the ChatGPT web and mobile apps. Enterprise users will gain access next week, broadening the model's availability across different user bases.