The intra-system biases, including differential code biases (DCBs) and differential phase biases (DPBs), are generally defined as receiver-dependent hardware delays between different frequencies in a single global navigation satellite system (GNSS) constellation. Likewise, inter-system biases (ISBs) are differential code and phase hardware delays between different GNSSs and are relevant for combined processing of multi-GNSS and multi-frequency observations. Although the two biases are usually assumed to remain unchanged for at least one day, they can exhibit remarkable intraday variability, likely due to environmental factors, particularly ambient temperature. We analyze whether the variability of the biases is sensitive to temperature and identify how this affects real-time kinematic (RTK) positioning. Tests using GPS, BDS-3, Galileo and QZSS observations collected by zero and short baselines show that ISBs for overlapping frequencies are fairly stable, whereas ISBs for non-overlapping frequencies can exhibit marked short-term variability driven by ambient temperature changes. By pre-calibrating and modeling these biases, empirical success rates and positioning performance can be significantly improved compared with classical and inter-system differencing models that assume time-invariant DCBs, DPBs and ISBs.