What’s New in WeatherMesh-6

Resources

Company

Contact<br>Sign In

Research<br>WindBorne<br>June 1, 2026

What’s New in WeatherMesh-6<br>By

Read the article<br>here.

The most skillful global weather model<br>WeatherMesh-6 (WM-6) Global is the most skillful medium-range forecasting model to date, whether NWP or AI. Operating at 0.25-degree (roughly 25km) resolution, it produces more accurate forecasts than the leading operational models—ECMWF's IFS and AIFS ensembles—across every weather parameter and lead time we evaluated. Over the evaluation window of July 2025 through March 2026, WM-6 achieves up to 38% lower ensemble-mean RMSE than IFS and up to 32% lower than AIFS. Most strikingly, for 2-meter temperature, WM-6's 4.5-day forecast is as accurate as a 1-day forecast from IFS. Operational forecast skill has historically advanced by roughly a day of lead time per decade; by that measure, WM-6 delivers in a single model what would have taken three decades.

Ensemble RMSE of WM-6 Global and AIFS relative to IFS. WM-6 Global has 128 ensemble members, while IFS and AIFS have 51 members. Lower RMSE = higher forecast accuracy. Since IFS Ensemble only forecasts out to 144 hours in their 6z and 18z runs, the first 144 hours are evaluated for all of 0z/6z/12z/18z runs and the rest are evaluated for 0z and 12z runs.Due to data access limitations, we were unable to directly compare WeatherMesh-6 with Google DeepMind's FGN, another AI weather model at the skill frontier. In indirect comparisons, WM-6 Global appears competitive with—if not slightly more skillful than—FGN across all variables we examined.<br>A more complete picture of the atmosphere<br>With 24 new surface, soil, and atmospheric parameters, in addition to 5 upper-air parameters at 25 pressure levels, WeatherMesh-6’s output catalog is the largest of any AI weather model to date . New soil temperature and moisture fields, along with skin temperature, give the model direct access to land-atmosphere coupling, fundamentally improving its ability to learn surface physics. A full suite of radiation variables enables downstream solar energy applications, while additions like cloud cover decomposed into high, medium, and low layers, and boundary layer height broaden WM-6's utility across sectors, from aviation to energy to agriculture. For a complete list of available parameters, see our API documentation.

24 new weather variables in WeatherMesh-6, shown from ERA5 reanalysis on July 15, 2024 at 12:00 UTC. Top row: skin temperature, soil temperature (layers 1 and 2), soil moisture (layers 1 and 2). Second row: surface pressure, total column water vapor, CAPE, boundary layer height, cloud base height. Third row: high, medium, and low cloud cover, snowfall, runoff. Fourth row: surface solar radiation downwards, surface thermal radiation downwards, top-of-atmosphere net solar radiation, surface net solar radiation, direct solar radiation. Bottom row: sensible heat flux, latent heat flux, forecast surface roughness, mean total precipitation rate, precipitation type (not in operational WM-6).‍<br>A more calibrated ensemble, in latent space<br>WeatherMesh-6 models a full ensemble in latent space—an approach we will detail in an upcoming technical report. This is reflected by a competitive CRPS (Continuous Ranked Probability Score), which captures both forecast skill and ensemble calibration. In our preliminary evaluation for the month of July 2025, WM-6's 128-member ensemble outperforms AIFS's 51-member ensemble on CRPS across key surface and atmospheric variables, and at every lead time from the initial hours out to 15 days.

CRPS of WeatherMesh-6 compared to AIFS Ensemble. Lower is better. WM-6 also outperforms AIFS on “fair” CRPS, which eliminates the effect of the difference in ensemble size. We show plain CRPS here because the 128-member ensemble is what we deliver in our operational forecasts.This approach also lets us produce realistic ensemble members. Each member is at full resolution and physically coherent across weather parameters , which enables sophisticated use cases such as scenario analysis and extreme-event detection, as well as parameter-coupled use cases like wind-energy forecasting. Alongside individual members for key parameters, we also distribute processed statistics, including percentiles and precipitation exceedance probabilities , so users can still understand the distribution of uncertainty if they do not wish to ingest the whole ensemble.

Top left: 360 h (15-day) deterministic forecast of total column water vapor, initialized 2025-08-23 12:00 UTC, valid 2025-09-07 12:00 UTC. Bottom left: 360 h ensemble-mean total column water vapor for the same init and valid time. Right: ensemble probability of 3-hourly precipitation exceeding 0.25 mm over the contiguous US, initialized 2025-08-25 18:00 UTC, valid 2025-08-28 21:00 UTC (lead time 75 h).‍<br>Improved data assimilation<br>With WeatherMesh-6, WindBorne’s AI DA architecture has been redesigned around ensemble methods :...