In early August, both Google and OpenAI slashed prices on their budget-friendly text-generating models, with Google reducing input costs by 78% and output by 71%, while OpenAI halved the input costs for GPT-4o.
According to Andy Thurai, principal analyst at Constellation Research, 'We expect the pricing pressure to continue with all AI models if there is no unique differentiator.' This underlines how competition shapes pricing dynamics.
John Lovelock, VP analyst at Gartner, highlights that models were initially priced to recover training costs, noting that OpenAI's GPT-4 training cost $78.4 million, but that competitive pricing is now essential.
Techniques like prompt caching and batching are used by vendors such as Google, Anthropic, and OpenAI to yield additional savings, showcasing the drive towards efficient operational practices within AI.
Collection
[
|
...
]