OpenAI reveals cheaper mini version of its flagship GPT-4o

19 Jul 2024

Image: © Catherine Zibo/Stock.adobe.com

OpenAI is advertising GPT-4o mini as a model that can perform various tasks at a low cost, as the AI sector continues to shift into developing small language models.

OpenAI has released a more cost-efficient version of its flagship GPT-4o model to reach a wider audience and expand how its products are used.

The company revealed a major GPT-4o upgrade earlier this year and said it will bring advanced features to its AI products, such as giving ChatGPT the ability to respond faster to text, audio and images – though some upgrades are facing delays.

But instead of power and performance, OpenAI is advertising GPT-4o mini as a model that can perform various tasks at a low cost and latency. The company also claims it is an upgrade from its previous “small models” such as GPT-3.5 Turbo. Users of ChatGPT’s Free, Plus and Team are now able to access GPT-4o mini.

“We envision a future where models become seamlessly integrated in every app and on every website,” OpenAI said in a blogpost. “GPT-4o mini is paving the way for developers to build and scale powerful AI applications more efficiently and affordably.”

OpenAI said it tested GPT-4o mini on “several key benchmarks” and claimed its model surpassed various other small models on the market “across both textual intelligence and multimodal reasoning”.

However, it’s important to take company claims about its own models with a pinch of salt. Last year, Google-owned DeepMind claimed that one of its AI models found 2.2m new crystals that could potentially be used to create new materials. This claim was later criticised by a group of researchers who said there was “scant evidence” the DeepMind AI found compounds that “fulfil the trifecta of novelty, credibility and utility”.

Earlier this year, the AI Index claimed that robust evaluations for large language models are “seriously lacking” and there is a lack standardisation in responsible AI reporting.

Large language models became the golden egg of the tech sector after the rise of ChatGPT, prompting many companies to develop their own generative AI models and applications.

But these models also have issues when it comes to both cost and energy use. As a result, some of the key players in the AI sector have been creating smaller, more accessible models. Google and Microsoft – two of the biggest players in the AI sector currently – have both made moves towards more lightweight AI options for their customers this year.

Find out how emerging tech trends are transforming tomorrow with our new podcast, Future Human: The Series. Listen now on Spotify, on Apple or wherever you get your podcasts.

Leigh Mc Gowran is a journalist with Silicon Republic

editorial@siliconrepublic.com