I’m excited to share @salientpredict’s newest generative AI weather model, GemAI v2. It delivers 200-member ensemble forecasts with a 126-day horizon and beats @ECMWF’s IFS ENS by a substantial margin across multiple variables and timescales.

The reason people hate Zohran is because he’s a spoiled retarded communist. His parents are privileged, placeless global citizens and Zohran was able to float along without real jobs in NYC taking a swing at a ‘rap career’ while living in a Chelsea apartment owned by his mother.

Third pass through Melissa. GoPro in side window as different camera looking forward shooting in ultra high res 8k. Not sure when that might get processed as the file turned out ridiculous. Barely had HD space for it and MacBook Pro promptly chocked when I tried to edit it

@WeatherMatrix @burgwx Here's a similar animation showing the probability of maximum daily wind gust (rather than mean surface level winds) above a 98th historical percentile threshold.

@GoogleDeepMind This was a long time in the making. Shoutout to @ViktorCikojevic, @LevangSam, @PauliusRauba, and everyone else who contributed to this release!

There’s a lot more to say about GemAI v2. The model is already operational and available through our GUI and API. We are planning to submit a paper on arXiv soon, along with a benchmark comparison against @GoogleDeepMind’s latest GenCast and FGN models

This year’s Gulf Coast blizzard sent natural gas prices soaring and stressed the Texas grid. Modeling this compound event requires accurately capturing the joint distribution of precipitation, wind speed, and temperature. GemAI’s physically realistic samples yield more

Visualizing forecasts this way highlights real-world impact. Here are the 2021 European floods that caused >$50 B in damages and claimed 200+ lives. GemAI accurately predicts the 3-day extreme precipitation event.

Since metrics never tell the full story, let’s look at a few notable extreme events, starting with Hurricane Beryl (2024). The animation illustrates how the probability forecast evolves day by day for a fixed target date. GemAI picks up the signal days before ENS.

Performance on extreme temperature, wind, and precipitation events, measured by the Brier Score, is also excellent. Note: metrics are computed using the latest 100-member ENS reforecast.

Here’s the money plot showing relative CRPS skill scores against the two leading NWP models, @ECMWF ENS and @NOAA GEFS. Scores are averaged across global landmasses over a 2-year reforecast period. Blue means Gem is better. We’re seeing 5–50 % improvement in skill depending on

GemAI is a generative vision transformer trained on 0.25° ERA5 reanalysis. It’s highly efficient: we can generate a single 126-day forecast sample on an H100 in ~20 seconds. It’s a natively probabilistic model, trained to optimize CRPS directly.

I'm giving a talk on ML for weather forecasting at #AMS2025. Feel free to stop by if you're attending the conference: https://t.co/n4qewWFRAr

I've consistently seen that organizations run by process fail, while organizations that win are run by a handful of opinionated individuals

This is an unprecedented attempt to stretch a law intended to apply in Europe to political activities in the US. It also patronizes European citizens, suggesting they are incapable of listening to a conversation and drawing their own conclusions.

It's difficult to overstate the danger inherent in the ideological alignment of the executive, its ostensibly independent agencies of intelligence and law enforcement, a largely captured institutional landscape, and a compliant media. Turnkey authoritarianism is not only

No.

Do not attempt to regulate what you do not understand

One reason you get more conservative as you age is because you come to learn that people don't really change; the unreliable, dishonest, disloyal, etc. tend to stay that way. You can't fix them, or save them, and this lesson applied at scale kills any tendency to utopianism.

https://t.co/6ImiAvHwp8