The globular cluster (GC) 47 Tuc has recently been proposed to host an intermediate-mass black hole (IMBH) or a population of stellar-mass black holes (BHs). To shed light on its dark content, we present an application of self-consistent multimass models with a varying mass function and content of stellar remnants, which we fit to various observational constraints. Our best-fitting model successfully matches the observables and correctly predicts the radial distribution of millisecond pulsars and their gravitational accelerations inferred from longterm timing observations. The data favours a population of BHs with a total mass of 430 +386 −301 M , but the most likely model has very few BHs. Since our models do not include a central IMBH and accurately reproduce the observations, we conclude that there is currently no need to invoke the presence of an IMBH in 47 Tuc. The global present-day mass function inferred is significantly depleted in low-mass stars (power-law slope α = −0.52 +0.17 −0.16 ). Given the orbit and predicted mass-loss history of this massive GC, the dearth of low-mass stars is difficult to explain with a standard initial mass function (IMF) followed by long-term preferential escape of low-mass stars driven by two-body relaxation, and instead suggests that 47 Tuc may have formed with a bottom-light IMF. We discuss alternative evolutionary origins for the flat mass function and ways to reconcile this with the low BH retention fraction. Finally, by capturing the effect of dark remnants, our method offers a new way to probe the IMF in a GC above the current main-sequence turn-off mass, for which we find a slope of −2.49 ± 0.08.