American research company OpenAI on Thursday (July 18), announced the launch of GPT-4o mini, its latest and most cost-effective small artificial intelligence (AI) model.
OpenAI says that it is over 60 per cent cheaper than GPT-3.5 Turbo and is already available as a text and vision model for developers, and ChatGPT web and app users.
Despite its affordability, GPT-4o mini boasts impressive performance metrics Scoring 82 per cent on the MMLU benchmark, it outperforms other small models like Gemini Flash and Claude Haiku. Its proficiency extends to mathematical reasoning and coding tasks, where it scored 87 per cent on MGSM and 87.2 per cent on HumanEval, respectively. Additionally, GPT-4o mini excels in multimodal reasoning, achieving a score of 59.4 per cent on MMMU, a multimodal reasoning evaluation.
GPT-4o mini supports both text and vision inputs in the API, with future plans to include image, video, and audio inputs and outputs.
ChatGPT Enterprise users can access the model next week. Additionally, fine-tuning capabilities for GPT-4o mini are expected to roll out in the coming days.
Comments