Open-weight models cross 40% of new enterprise AI deployments
Cost and control concerns are pulling enterprise buyers away from closed frontier models faster than expected.

Open-weight models now account for more than 40% of new enterprise AI deployments tracked by Wolfsberg Index, up from roughly a quarter a year ago.
Procurement teams cite data control, fine-tuning flexibility, and total cost of ownership as the leading reasons, ahead of raw capability comparisons.
Closed-model providers are responding with steeper enterprise discounts and more permissive fine-tuning terms, narrowing the gap on the factors buyers say matter most.
The shift is most pronounced in regulated industries — financial services and healthcare in particular — where the ability to run models entirely within a controlled environment addresses compliance requirements that closed, API-only models struggle to satisfy.
Capability gaps between open-weight and closed frontier models have also narrowed meaningfully over the past year, reducing what was previously the strongest argument for defaulting to closed models regardless of cost or control considerations.
Enterprise buyers who have already switched describe the transition as operationally heavier upfront — requiring more in-house infrastructure and tuning expertise — but consistently cheaper at scale, which is increasingly winning out as deployments move from pilot projects to production workloads.


