by NOAAUS
Give US forecasters global forecasts as good as or better than the physics GFS at a sliver of the computing cost, so more and faster runs fit on the same machines.
Operational since Dec 2025; v1.1 (July 2026) improved hurricane intensity and long-range blur.
Updated 27 Jul 2026Checked 11 Oct0 updates this week
The network learned how the global atmosphere changes from one step to the next by studying decades of past weather states.
Each run starts from NOAA's best estimate of the current atmosphere and steps forward repeatedly to 16 days, in about 40 minutes.
31 slightly different starting states give 31 forecasts; how much they disagree tells forecasters how sure to be.
HGEFS adds the 31 AI runs to the 31 physics runs; mixing two kinds of errors gives a better combined picture.
| Spec | AIGFS |
|---|---|
| Compute vs physics GFS (16 days) | 0.3%R (reported) |
| 16-day forecast run time | About 40 minutesR (reported) |
| AI ensemble members (AIGEFS) | 31R (reported) |
| AIGEFS compute vs physics GEFS | About 9%R (reported) |
| Hybrid ensemble members (HGEFS) | 62 (31 AI + 31 physics)R (reported) |
| AIGEFS extra useful lead time | 18 to 24 hours (early results)R (reported) |
| Base model | GraphCast (Google DeepMind), fine-tunedR (reported) |
R reported by the company
NOAA
US weather service, AI global forecast models
The US National Oceanic and Atmospheric Administration runs the National Weather Service and its forecast models. Since December 2025 it also runs AI forecast models alongside its physics-based GFS.
