The AI boom has beenbuilt ona basic assumption:Bigger models are more powerful, and the most powerful models win.Now, the industry is about to learn what happens ifthat assumption starts to break.
Mounting costs havealreadypressured users to give smaller and cheaper models a second look.Thiscost-conscious model-shoppingisnewandit’sunclear how it will affect the industry, but the impact is likely tobesignificant.
Oneprediction, laid out best by Coinbase co-founder Brian Armstrong,is that it will result in the vast majority of tasks shifting to cheaper models.
“[D]emand forintelligence is near infinite, but 80% of workloads will be running on 99% cheaper models within 12-18 months,” Armstrongwrote on X. “20% of workloads will still run on latest gen models where IQ maxing is important.”
It’shard to overstate what a significant shift it will be for the AI industry if Armstrong’s predictioncomes true.
Before now,most AI companies havecompetedon quality, which has meantdefaulting to the most advanced available model. If those same jobs can behandled bycheaper models without affecting quality, it would mean a massive shift in the economics ofAI.And critically, muchof the savings would be coming out of the pockets of the big labs,dealingafinancialblow to OpenAI and Anthropic just asthey’reheading fortheirIPOs.
It’sa potentially seismic change in theindustry, resting on one basic question:Are companiesreadyto switch to smaller models?
Initial tests suggest that, when the system is arranged right, cheaper models could sub in without any sacrifice in quality. In a recent test by the legal AI tool Harvey,thecompany was able to reduceinference costsby 3xwithout reducing quality. Thetest, performed in partnershipwith the inference platform Fireworks AI, combinedClaude Opus andFireworks’GLM 5.1,and shifted toOpus for the most intensive tasks.The result was a significantly lower load in terms of server time and overall cost.
“Quality comes first, and in legal it always will,” Harvey co-founder Gabe Pereyra told TechCrunch, referring to the AI legal services his startup provides. “However, the definition of quality is evolving from simply using the most powerful model for everything, to using the best model that gets the right answer most efficiently.”
Thistrendis often framedin terms ofmajor labs versusChinesemodelsoropen-weightones,but that misses the bigger point. The real divideisn’tbetweenproprietary and open models;it’sbetweenlarge models and small ones.Youcan save money by switching from GPT-5.5 to DeepSeek’s V4Flash, butswitching toGPT-5.4-mini works just as well.
There’san active price war going on between in-house inference from the big labs and independently served open-weight models. For the bigger question of small versuslarge, itdoesn’treally matter which kind of small model wins out.
Allofthis might seem obvious — of course youshouldn’tuse morecomputethannecessary— but itruns counter tothe scaling-first approach that has dominated the industry until now. Inspired bythebitter lessonlabs have leaned hard into training the most compute-intensive models possible,pushingthe frontier of what AI modelscando. With prices heavily subsidized by investors, clients had no reason to choose anything but the most advanced option.
With token prices rising and subsidies slowing down, users are facing cost pressure for the first time.We don’t know whether the new cost pressure will actually drive enterprise users to smaller models.They could just as easily economize by making fewer calls, using lesscontext,or simply giving up on the least promising deployments.
But if it turns out that most deployments can be run just as well on a smaller model, it couldput aserious damper on the growing demand for inference — and raise new questions about how to justify the cost of training a frontier model.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.