AbstractOver recent years there has been increasing interest from commercial companies to produce observations of the atmosphere using GNSS‐RO (Global Navigation Satellite System – Radio Occultation) techniques. One such company, Spire, has achieved this by launching a large number of cubesats that operate in a constellation to observe the atmosphere. Over a period of 18 months, these data have been assessed in comparison to current GNSS‐RO data. The differences between the observations and a short‐range forecast (the innovations) indicate that the data are closely comparable to currently operational GNSS‐RO data below 30 km. Above this height, the data are noisier than current satellites. Below 30 km the data often have a smaller standard deviation than other observations, which can be achieved by using increased vertical smoothing. Although the vertical correlation length‐scales for Spire data are larger than for the current data, they do not seem excessively large. Given the good performance of the innovation statistics for the Spire data, impact trials were run in the Met Office's NWP system. Tests were run assimilating data from the current observational network, and with no GNSS‐RO data, adding data from Spire in various proportions, and replacing the data from the Metop‐C satellite with data from Spire. It was found that there is a substantial forecast benefit from assimilating Spire data. In accordance with previous work, the forecast impact of increasing the data volume is roughly proportional to the logarithm of the total amount of GNSS‐RO data assimilated. Theoretical arguments suggest a different relationship with the number of observations, but these do not fit the results well.