OpenAI today announced a reduced-price “mini” model that it says will enable more companies and programs to take advantage of its artificial intelligence. The new model, called GPT-4o mini and available today, is 60 percent cheaper than OpenAI’s existing cheapest model while still offering increased performance, the company says.
OpenAI describes the move as part of an effort to make AI “as accessible as possible,” but it also reflects increasing competition among cloud AI providers, as well as growing interest in small, free, open-source AI models. Meta is expected to release the largest version of its very capable free offering, Llama 3, next week.
“OpenAI’s goal is to create and distribute AI securely and make it widely accessible,” Olivier Godement, OpenAI’s product manager responsible for the new model, tells WIRED. “Making intelligence available at a lower cost is one of the most efficient ways to do that.”
Godement says the company developed a more cost-effective offering by improving the model architecture and refining the training data and training regimen. GPT-4o mini outperforms other “small” models on the market on several common benchmarks, OpenAI says.
OpenAI has carved out a significant niche in the cloud AI market thanks to the extraordinary capabilities of its chatbot, ChatGPT, which debuted in late 2022. The company allows outside users to access the large language model that powers ChatGPT, called GPT-4o, for a fee. It also offers a less powerful model, called GPT-3.5 Turbo, for about a tenth of the cost of GPT-4o.
The interest in language models sparked by ChatGPT’s huge success has spurred competitors to develop similar offerings. Google, a pioneer in artificial intelligence, has made a major push to create and market a large language model and chatbot under the Gemini brand. Startups such as Anthropic, Cohere, and AI21 have raised millions to develop and market their own large language models for commercial customers and developers.
Creating the highest-performing large language models requires huge financial resources, but some companies have chosen to open source their creations to attract developers to their ecosystems. The most prominent open-source AI model is Meta’s Llama; it is free to download and use, but its license imposes certain limits on commercial use.
In April, Meta announced Llama 3, its most powerful free model. The company released a small version of the model with 8 billion parameters (a rough measure of a model’s portability and complexity), as well as a more powerful, mid-sized version with 70 billion parameters. The mid-sized model is close to OpenAI’s best offering on several benchmark scores.
Multiple sources confirmed to WIRED that Meta plans to release the largest version of Llama 3, with 400 billion parameters, on July 23, though they say the release date could change. It’s unclear how capable this version of Llama 3 will be, but some companies have turned their attention to open-source AI models because they’re cheaper and more customizable, and offer greater control over a model and the data fed to it.
Godement recognizes that customer needs are evolving. “What we are seeing more and more in the market is developers and enterprises combining small and large models to create the best product experience at the price and latency that suits them,” he says.
Godement says OpenAI’s cloud offerings offer customers models that have gone through more security testing than competitors. He adds that OpenAI could eventually develop models that customers can run on their own devices. “If we see massive demand, we can open that door,” he says.