OpenAI has introduced its latest small AI model, GPT-4o Mini, which promises to revolutionize the AI landscape by offering a more affordable and faster alternative to existing models. Released on Thursday, GPT-4o Mini is now available for developers, as well as for consumers through the ChatGPT web and mobile app. Enterprise users will gain access next week. This new model aims to make AI more accessible and cost-effective, aligning with OpenAI’s mission of democratizing AI technology.
Performance and Capabilities
GPT-4o Mini is designed to outperform industry-leading small AI models in reasoning tasks involving text and vision. It replaces GPT-3.5 Turbo as the smallest model offered by OpenAI and is positioned as a superior alternative to models like Google's Gemini 1.5 Flash and Anthropic’s Claude 3 Haiku. The new model achieves impressive scores of 82% on the Measuring Massive Multitask Language Understanding (MMLU) benchmark and 87% on the MGSM math reasoning test, surpassing its competitors in both metrics.
Cost Efficiency and Speed
Major advantages of GPT-4o Mini are its cost efficiency and it's 100x cheaper than OpenAI's GPT-3 Davinci model It is significantly more affordable to run than its predecessors, with a runtime cost of more than 60% lower than GPT-3.5 Turbo. For developers using OpenAI’s API, GPT-4o Mini is priced at 15 cents per million input tokens and 60 cents per million output tokens, making it an attractive option for high-volume, simple tasks. The model also boasts a context window of 128,000 tokens, equivalent to the length of a book, and has a knowledge cutoff of October 2023.
In terms of speed, GPT-4o Mini delivers a median output speed of 202 tokens per second, more than twice as fast as GPT-4o and GPT-3.5 Turbo. This rapid processing capability makes it ideal for speed-dependent use cases, including many consumer applications and agentic approaches to using large language models (LLMs).
Multimodal Support and Future Enhancements
Currently, GPT-4o Mini supports text and vision in the API, but Sam Altman confirms on X to release two models in a month Alpha and GA including voice capabilities. This multimodal functionality could enable more advanced virtual assistants that understand and respond to complex user needs, such as creating travel itinerary suggestions based on uploaded documents and images.
Enterprise Tools and Compliance
In addition to GPT-4o Mini, OpenAI announced new tools for enterprise customers. The Enterprise Compliance API is designed to help businesses in highly regulated industries, such as finance, healthcare, legal services, and government, comply with logging and audit requirements. This API will provide records of time-stamped interactions, including conversations, uploaded files, and workspace users, giving admins greater control and transparency over their ChatGPT Enterprise data.
Moreover, OpenAI is enhancing its granular control features for workspace GPTs, as well as custom versions of ChatGPT tailored for specific business use cases. Previously, admins could only fully allow or block GPT actions within their workspace. Now, they can create an approved list of domains that GPTs can interact with, providing more nuanced control over the AI's functionality.