هزینه هوش مصنوعی در حال سقوط است

This isn't as strong as GPT-6 Astra but it's far more efficient and it gets to the heart of where AI is headed. AI is becoming a commodity faster than it is becoming an indispensable service.

Already we have the Sol model, which is near Astra at a fifth of the price but now we have Terra, which cuts the price in half again. OpenAI also touts the model as being particularly useful for automated computer use.

"Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it." I'm skeptical of the 'writes naturally' claims from any of the models but there has been steady improvements across all tasks. For markets, falling costs make adoption more attractive, but they also raise questions about the margins and returns on capital of the companies supplying AI infrastructure.

Ultimately, I strongly believe that all the benefits of AI will fall to incumbent asset-heavy companies that can use it to improve efficiency rather than model makers. Now the bulls will argue Jevon's Paradox and I think that can run for awhile (years) but models are getting so intelligent and cheap that soon they will have extremely cheap models that cover 99% of use cases.

Along that curve, the profits will shift from tech companies to productivity beneficiaries, with layoffs along the way. Think about a company with substantial labour expenses and relatively limited exposure to AI-related capital spending. If it can automate 10–20% of routine administrative work, the potential margin impact could be enormous.

As for users, here is a better look at the somewhat-confusing lineup of models that OpenAI is currently offering.

The Nasdaq was last up 1.0% and the big news today is that Anthropic has filed for an IPO.

This article was written by Adam Button at investinglive.com.