OpenAI Releases GPT-4o Mini, New AI Model At Cheaper Prices

JAKARTA - The company behind ChatGPT, OpenAI announced that it has launched a mini GPT-4o, its latest smaller artificial intelligence model and also the most cost-effective.

The company hopes the mini GPT-4o can expand the reach of applications built with AI significantly by creating much more affordable intelligence.

Currently, the mini GPT-4o supports text and visuals in the API, with support for text input and output, images, videos, and audio to come in the future.

The model is also said to have a context window of 128K tokens, support up to 16K output tokens per request, and have knowledge until October 2023.

"Thanks to the enhanced tokenizer shared with GPT-4o, non-English text handling is now becoming more cost-effective," the company said in its blog quoted Friday, July 19.

Although smaller, OpenAI says that the mini GPT-4o surpassed GPT-3.5 Turbo and other small models on academic benchmarks across multimodal textual intelligence and reasoning, and supports the same language range as GPT-4o.

The model also shows a strong performance in functional summons, which can allow developers to build apps that take data or take action with external systems, compared to GPT-3.5 Turbo.

No need to worry about security, as OpenAI ensures that mini GPT-4o has the same safety mitigation already installed such as GPT-4o.

"Based on this learning, our team is also working to improve the security of the mini GPT-4o using a new technique informed by our research," he added.

On ChatGPT, Free, Plus, and Team users will be able to access mini GPT-4o starting today, replacing GPT-3.5. Meanwhile, Enterprise users will also have access starting next week.