Weather is a central determinant of production risk in agricultural markets, but markets respond not only to weather conditions themselves, but also to how those conditions are noticed, framed, and interpreted in public discourse. In this setting, real-time online weather discussions may shape price uncertainty by influencing how market participants process evolving production risk. This study examines whether daily weather-related online media attention and sentiment contain information relevant for short-run conditional variance in U.S. corn, soybean, and winter wheat futures. Using 30.6 million weather-related online mentions from January 2022 through January 2025, we construct commodity-specific indicators of weather mentions and sentiment and incorporate them into GARCH models alongside key USDA report dates. The results reveal economically meaningful links between online weather discourse and short-run variance dynamics, though their strength and consistency vary across markets. In corn, increases in weather-related mentions are associated with higher conditional variance, while more positive sentiment is generally associated with reductions in conditional variance. Soybeans show smaller but consistently stabilizing variance responses. Wheat exhibits fewer significant media effects, but when present they imply economically large proportional changes in the heteroskedastic component of conditional variance, on the order of 40%–60% relative to baseline. Taken together, the results indicate that weather-related online media provide incremental information about short-run price uncertainty in agricultural futures markets. USDA reports continue to generate the largest discrete variance adjustments, but weather-related attention and sentiment help explain additional, higher-frequency changes in conditional variance between scheduled releases. In this sense, online weather discourse appears to reflect how developing production concerns are interpreted in real time, offering a complementary signal of evolving market uncertainty.