Normal-mode splitting reveals coherent cavity–magnon hybridization but does not certify finite-time quantum-state transfer. We describe quantum transfer in a passive, linear cavity–magnon system coupled to thermal reservoirs as a thermal-loss channel specified by its coherent transmissivity η and output-referred thermal population ν. For single-rail qubits, we calculate the phase-corrected average fidelity and compare it with the deterministic measure-and-prepare benchmark of 2/3. A Bell pair shared by the cavity and an isolated reference retains reference–magnon entanglement if and only if η>ν. An input squeezed vacuum of strength r yields sub-vacuum magnon fluctuations if and only if ν<η(1-e⁻²ʳ)/2. When the initial magnon and both reservoirs have the same occupation, an analytically determined peak-transmissivity time optimizes all three tasks. These task-dependent benchmarks show that a conventional linewidth-based strong-coupling criterion does not guarantee quantum-transfer performance. Conversely, entanglement or squeezing can survive below that spectroscopic reference when the added noise is sufficiently low. These benchmarks provide quantitative targets for transfer timing and thermal-noise control in intracavity-to-magnon quantum interfaces.