OpenAI's GPT-4o mini is now reportedly available for consumers and developers. It is the latest mini artificial intelligence model that prioritizes efficiency and speed.
According to the business, GPT-4o mini performs better than industry-leading tiny AI models on reasoning tasks, including text and visual.
Because small AI models are faster and less expensive than larger models like GPT-4 Omni or Claude 3.5 Sonnet, they are gaining popularity among developers as they improve. They are a good choice for frequent, straightforward tasks that developers may ask an AI model to complete.
(Photo : SEBASTIEN BOZON/AFP via Getty Images)
A new study reveals that using OpenAI's ChatGPT for scriptwriting could impose major limitations on artistic creativity.
The GPT-4o mini will replace the GPT-3.5 Turbo as OpenAI's most minor variant. According to statistics from Artificial Analysis, the company's most recent AI model scored 82% on the MMLU benchmark, which measures reasoning, as opposed to 79% for Gemini 1.5 Flash and 75% for Claude 3 Haiku.
GPT-4o mini earned 87% on the MGSM, a test of math thinking, while Flash and Haiku got 78% and 72%, respectively.
Additionally, OpenAI claims that the GPT-4o mini is more than 60% less expensive to operate than the GPT-3.5 Turbo and much more affordable to run than its prior frontier models.
The GPT-4o mini model now supports text and vision through the API, but according to OpenAI, video and audio capabilities will be added.
Read Also: OpenAI Nearing Level Two of Five in Building an AI Model that Outperforms Humans
GPT-4o Mini Costs
GPT4o mini costs 15 cents for each million input tokens and 60 cents for each million output tokens for developers using OpenAI's Java Programming Interface. The model's knowledge cutoff is October 2023, and its context window is 128,000 tokens long or nearly a book length.
OpenAI stated that GPT-4o mini is around the same size as other small AI models, such as Llama 3 8b, Claude Haiku, and Gemini 1.5 Flash, but would not specify the size.
However, according to pre-launch testing conducted in the LMSYS.org chatbot domain, the business asserts that the GPT-4o mini is quicker, more economical, and more intelligent than leading miniature versions in the sector. Independent testing conducted early on supports this.
Mini AI Model Prominence
OpenAI's GPT-4o mini is the latest model after numerous AI giants have released their cost-effective AI models.
In April, Microsoft unveiled Phi-3-mini, a model that the tech giant boasts is a "dramatically cheaper" AI.
Microsoft's vice president of GenAI research, Sébastien Bubeck, claims that Phi-3 is somewhat more affordable than other models with similar capabilities and much less expensive.
Microsoft asserts that the Phi-3 family of open models is the most efficient and cost-effective microlanguage model, even though large language models are still the "golden standard" for various AI applications.
It is claimed that Phi-3 models perform better on a range of benchmarks that assess language, coding, and mathematics abilities than models of the same size and the next level up.
The Microsoft Azure AI Model Catalog, Hugging Face, a platform for machine learning models, Ollama, a lightweight framework for running models locally, and other places can all offer the new lightweight alternative.
Related Article: Grok AI Delivers Near-Instantaneous Responses, Outpacing Common Models
