Introducing the GPT-4o mini, a revolutionary small model from OpenAI designed to make advanced AI intelligence more accessible and affordable. This model scores an impressive 82% on the MMLU benchmark, outperforming even the GPT-4 in chat preferences, as evidenced by its position on the LMSYS leaderboard. Priced at just 15 cents per million input tokens and 60 cents per million output tokens, GPT-4o mini offers an order of magnitude more affordability compared to previous models, making it over 60% cheaper than GPT-3.5 Turbo.

GPT-4o mini is engineered to support a wide array of applications due to its low cost and latency. It is particularly effective in scenarios that require multiple model calls, handling large volumes of context, or providing real-time text responses, such as in customer support chatbots. Currently, the model supports text and vision inputs and outputs through its API, with plans to extend capabilities to include image, video, and audio in the future. It features a context window of 128K tokens and can output up to 16K tokens per request, with knowledge updated up to October 2023.

One of the standout features of GPT-4o mini is its superior performance in both textual intelligence and multimodal reasoning. It surpasses other small models like GPT-3.5 Turbo and Claude Haiku on various academic benchmarks. The model excels in reasoning tasks involving text and vision, mathematical reasoning, and coding proficiency. For instance, it scores 87.0% on MGSM for math reasoning and 87.2% on HumanEval for coding performance, significantly outperforming its competitors.

Safety and reliability are built into GPT-4o mini from the ground up. OpenAI has implemented rigorous safety measures, including filtering out harmful content during pre-training and using techniques like reinforcement learning with human feedback (RLHF) for post-training alignment. The model incorporates the same safety mitigations as GPT-4o, ensuring it is safe for use in various applications.

GPT-4o mini is now available in the Assistants API, Chat Completions API, and Batch API. It is priced affordably, making it accessible to a broad range of users, including those on Free, Plus, and Team plans in ChatGPT, as well as Enterprise users. The introduction of GPT-4o mini marks a significant step towards making AI more integrated into everyday digital experiences, offering developers the ability to build and scale powerful AI applications more efficiently and affordably.

In summary, GPT-4o mini is a groundbreaking model that combines superior performance, multimodal capabilities, and robust safety features with unprecedented affordability. It is poised to revolutionize the way AI is used in various applications, making advanced intelligence more accessible to everyone.