OpenAI, Meta, SpaceXAI release new AI models touting cost efficiency

The competition among leading artificial intelligence developers has entered a new phase — one defined not by which model is smartest, but by which is cheapest to use. Over the past week, OpenAI, Meta ↘ 0.39%, and Elon Musk's SpaceXAI all released new models whose primary selling point is cost efficiency, a shift that reflects growing unease among business customers over ballooning AI bills.

Sticker Shock Drives the Pivot

OpenAI said its latest offering, GPT-5.6, is designed to complete more work while using significantly fewer tokens. SpaceXAI's Grok 4.5 claims twice the token efficiency of comparable models. And Meta CEO Mark Zuckerberg told Bloomberg that pricing for the company's Muse Spark 1.1 will be "attractive," adding that "there's a real ability to be able to offer frontier or very high-level intelligence at a much more affordable cost."japantimes+1

The shift follows months of companies grappling with unexpectedly large AI invoices. "Companies are spending a lot more than they used to," said Gil Luria, head of technology research at DA Davidson. "As they see these costs get out of control, they're starting to ask questions about efficiency."moneycontrol

Gautier Cloix, CEO of Paris-based AI startup H Company, said one executive showed him a monthly AI usage invoice running into the millions of dollars. Some firms have imposed tighter spending limits after a period of encouraging employees to use AI as much as possible — a practice known as "tokenmaxxing."moneycontrol

From Power to Price

OpenAI CEO Sam Altman acknowledged the changed environment in an interview with CNBC last week. "Every enterprise now is thinking about spend and the value they're getting in exchange for AI, and this is what we really want to do," he said. The rhetoric marks a departure from a year ago, when OpenAI executives publicly discussed one day charging thousands of dollars per month for top-tier subscriptions.moneycontrol

The cost push also puts pressure on Anthropic, whose Opus and Fable models rank among the most expensive on a cost-per-task basis, according to benchmarking service Artificial Analysis. Musk took direct aim at the rival, writing on social media that Grok 4.5 "is an Opus-class model, but faster, more token-efficient and lower cost."moneycontrol

A Bullish Signal for Investors

Investor Gavin Baker, founder of Atreides Management, has framed the falling cost of intelligence as among the most bullish developments for the AI ecosystem. In a recent podcast appearance, Baker argued that declining token costs drive exponential usage growth, noting that the shift to usage-based pricing could push OpenAI's annualized revenue well above $200 billion this year. Chinese competitors like DeepSeek continue to offer cheaper open models, while model-routing services such as OpenRouter — which raised more than $100 million in May — are helping businesses shop for the best price across hundreds of models.podcasttranscript+2