US-based artificial intelligence firm OpenAI announced the launch of a new generative AI model dubbed “GPT-4o mini” on July 18.
According to a blog post from OpenAI, the new model is “an order of magnitude more affordable than previous frontier models,” and “more than 60% cheaper than GPT-3.5 Turbo.”
Performance vs power
GPT-4o mini is essentially the cost-effective version of the current top-of-the-line consumer model for OpenAI’s flagship product, ChatGPT.
Per OpenAI, the tradeoff between power and performance are minimal. Despite operating with a much smaller energy consumption footprint, GPT mini doesn’t appear to be lacking.
The company writes:
“GPT-4o mini surpasses GPT-3.5 Turbo and other small models on academic benchmarks across both textual intelligence and multimodal reasoning and supports the same range of languages as GPT-4o.”
GPT Mini
The new model supports much of the same functionality as its predecessor, however it is currently limited to text and vision. OpenAI says support for audio and video are coming soon.
It’s unclear at this time if the mini model has any environmental benefits over other models. OpenAI hasn’t released information pertaining to the actual methodology used to reduce running costs. This could indicate the benefits don’t extend to actual energy savings but instead may only apply to end-user cost savings.
OpenAI at large
The launch comes amid a steady stream of activity from the company and no shortage of actions against it.
As Cointelegraph recently reported, OpenAI is allegedly working on an AI model capable of advanced reasoning relative to GPT-4o. Dubbed “Strawberry,” the new model will purportedly be capable of more humanlike responses. Reportedly, Strawberry is an extension of the company’s mysterious Q* project.
At the other end of the news spectrum, OpenAI could end up in the crosshairs of the Securities and Exchange Commission after whistleblowers called on the agency to investigate potential wrongdoing relative to the company’s use of non-disclosure agreements.