OpenAI, the artificial intelligence startup backed by Microsoft, has unveiled its latest AI model, “GPT-4o mini,” during the Microsoft Build conference held at the company’s headquarters in Redmond, Washington. The release of this new model marks OpenAI’s ongoing efforts to enhance the functionality of its widely-used chatbot. Described as the “most capable and cost-efficient small model available today,” GPT-4o mini is expected to incorporate image, video, and audio capabilities in the future.
GPT-4o mini is derived from OpenAI’s most advanced model, GPT-4o, which was introduced in May during a livestreamed event with company executives. The “o” in GPT-4o stands for omni, highlighting the model’s improved audio, video, and text capabilities. OpenAI claims that GPT-4o can handle 50 different languages with enhanced speed and quality.
With a valuation exceeding $80 billion, OpenAI faces the challenge of maintaining its dominance in the generative AI market while seeking profitable avenues amidst substantial investments in processors and infrastructure for model development and training. The introduction of GPT-4o mini aligns with OpenAI’s strategic focus on “multimodality,” aiming to offer a wide range of AI-generated media, including text, images, audio, and video, within a single tool called ChatGPT.
OpenAI’s COO, Brad Lightcap, emphasized the importance of multimodality, stating that the world is not limited to text and code as the sole interfaces for powerful AI models. He believes that incorporating various modalities is essential to replicate how humans process and engage with the world.
Starting from Thursday, GPT-4o mini is available to free users of ChatGPT, as well as ChatGPT Plus and Team subscribers. It will be accessible to ChatGPT Enterprise users from next week, according to the company’s official release.