Gartner analysts Mark Petty and Raymond Paquet say “nine in 10 agent workflow steps are routine,” The Register reports, and open-weight models handle them “at parity for a quarter of the price of the cheapest closed model.”
The claims
- Against a frontier model on every step, open models cost “as little as a twentieth”
- In practice they’re wordier and retry more, so the real saving is closer to 5-6x
- Prediction: by 2030, 90% of enterprise operational agent workflow steps will run on open-weight models, mostly domain-specific ones
- The value is in the harness that routes each step to the right model
Their best line: “Open-weights preserve your control. The cost saving is a bonus.”
My take
That matches the local-AI case: send the routine stuff to a small model you control and save the frontier model for the hard parts.
God Bless, and May the Source Be with You!
