The use of global navigation satellite systems (GNSS) is a competitive way to provide high-precision and low-cost time and frequency transfer results. Traditional precise point positioning (PPP) is usually based on the ionosphere-free combination, which is not flexible in multi-frequency scenarios, and it relies on precise satellite clock products that limit time and frequency transfer performance. Achieving integer ambiguity resolution (IAR) is also challenging. In this contribution, a new undifferenced and uncombined (UDUC) GNSS time and frequency model is derived. Satellite clocks are estimated with other parameters, and integer ambiguities are resolved in double-differenced form. Numerical tests show that, with ambiguities resolved, the UDUC model with fixed satellite clocks improves by 20%-50% compared with UDUC PPP; with IAR and estimated satellite clocks, the proposed model improves by 10%-40% over the fixed-clock model; and GPS, Galileo, and BDS-3 all have the potential to achieve frequency transfer in the low-to-mid 10^-17 range for averaging times within one day.