The instinct is to reach for the biggest model. In the enterprise, that instinct is usually wrong.
Leverage comes from fit, not size
A model fine-tuned on your domain, wrapped in a good evaluation harness and served cheaply, will out-earn a frontier API you can't afford to run at scale. The question is never "what's the best model?" — it's "what's the best model for this task, at this cost, with these guarantees?"
Where open weights win
- Cost predictability — you own the curve.
- Data residency — the model runs where your data has to stay.
- Control — you can change it, quantize it, and audit it.
The frontier is a wonderful place to prototype. Production rewards the boring, controllable choice.