▲ OpenAI
Bloomberg reported that AI developers OpenAI, Meta Platforms (Meta), and SpaceX AI have recently launched a series of low-cost, high-efficiency AI models, intensifying the competition for cost-effectiveness.
The analysis suggests that as the AI transformation boom leads to a surge in corporate AI usage fees and low-cost Chinese AI models become increasingly competitive, it is becoming difficult for industry leaders to maintain a premium strategy that emphasizes performance alone.
According to Bloomberg, OpenAI, Meta, and SpaceX AI each unveiled practical, cost-effective AI models last week.
OpenAI’s new GPT-5.6 focuses on enabling more tasks while consuming fewer tokens (a unit of AI usage) than previous versions.
SpaceX AI also launched its Grok 4.5 model, highlighting that its token efficiency is twice that of comparable AI from competitors as a key selling point.
Meta CEO Mark Zuckerberg told Bloomberg that the company’s new Muse Spark 1.1 model will also focus on strengthening price competitiveness.
As major companies rapidly adopt AI agents—advanced task assistants—across various fields such as coding, content creation, research and development (R&D), and production management, AI usage fees have become a significant burden on corporate financial structures.
AI developers like OpenAI typically apply a pay-as-you-go system that charges fees in proportion to the number of tokens used.
At the same time, as many client companies engage in "tokenmaxxing"—drastically increasing their internal AI usage to compete in AI adoption—the financial burden of AI has soared to unsustainable levels.
Gauthier Clua, CEO of the French AI startup H Company, said in an interview with Bloomberg, "One CEO showed me their AI bill, and it amounted to millions of dollars per month."
Consequently, improving cost-effectiveness has become a core task for top-tier AI developers like OpenAI.
Simply maintaining a lead in AI performance through massive investments in AI infrastructure and R&D is no longer enough to retain customers.
OpenAI is cited as the most significant example of this shift.
Just a year ago, management confidently declared that they could continue to raise corporate subscription fees due to the superiority of their advanced AI models, but recently, they have been mentioning the need to secure price competitiveness much more frequently.
OpenAI CEO Sam Altman stated in a recent interview with CNBC, "Every company is currently weighing the cost burden against the practical value of AI adoption," adding, "Helping companies feel the value relative to the cost is our core objective."
Last month, OpenAI introduced an analysis tool to help its customers monitor AI usage and updated features to manage spending caps.
Meta, a late entrant to the corporate AI market, has emphasized targeting the market with practical, cost-effective AI from the start.
CEO Zuckerberg argued, "The pricing policies of some AI developers are extremely expensive and involve excessive profit margins," adding, "We have the capability to provide top-tier AI at a much more reasonable cost."
Bloomberg noted that this competition for cost-effectiveness could also put pressure on Anthropic, a major player in the corporate AI market, to lower its prices.
According to market research firm Artificial Analytics, Anthropic’s flagship models, Opus and Fable, are currently classified as the most expensive AI product lines in the industry.
(Photo: AP, Yonhap News)
※
Copying, redistribution, and unauthorized use in AI training are strictly prohibited.