NOAA GFS (US National Oceanic and Atmospheric Administration) and ECCC GEM (Environment and Climate Change Canada) are two of the models behind the forecast you see in almost every weather app. We grade both against the weather that actually happened, so this is a measured comparison rather than an opinion.
Short answer
American and Env. Canada are evenly matched across the 1 market we grade. Both are graded against ERA5 reanalysis over a rolling 30-day window, scoring temperature, wind and rain. No model wins every day or every place, which is why we show them side by side.
Head to head, by market
CheckWeather Score — measured against recorded weather
Market
GFS
GEM
Winner
Canada300 cities
75±1.2°C temp · 77% rain
75±1.61°C temp · 82% rain
Tie
Scores blend temperature, wind and rain accuracy, each measured against ERA5 (the recorded weather) over a rolling 30-day window. Full method.
How they differ
same physics, different machines
American · GFS
The United States' flagship global model, fully open and updated four times a day. It powers a huge share of the free weather apps and tools on the internet.
Run by US National Oceanic and Atmospheric Administration Resolution ~13–25 km
Env. Canada · GEM
Canada's national model, built for a country of extreme cold snaps and fast-moving systems. Published as open data by Environment Canada.
Run by Environment and Climate Change Canada Resolution ~15 km global
Common questions
Is GFS more accurate than GEM?
In our own testing, the two are close to level across the 1 market where both are graded. We measure each model's archived forecast against ERA5 reanalysis (the recorded weather) over a rolling 30-day window, scoring temperature, wind and rain. No model wins everywhere or every day, which is exactly why we show both side by side rather than picking one.
What is the difference between GFS and GEM?
NOAA GFS is run by US National Oceanic and Atmospheric Administration (United States) at roughly ~13–25 km resolution. ECCC GEM is run by Environment and Climate Change Canada (Canada) at roughly ~15 km global. They solve the same atmospheric physics with different code, different resolution and different data assimilation, which is why they disagree - and that disagreement is a useful signal about how certain the forecast really is.
Which should I trust for my forecast?
Neither on its own. When GFS and GEM agree, the forecast is reliable and you can plan on it. When they split, the day is genuinely uncertain and any single number is hiding that from you. That is why every city page on checkweather.io shows both models with a plain-English confidence score instead of one blended figure.