The engineering stress-strain curve is often observed to be flat across a wide range of plastic strains after yielding in a uniaxial tensile test, for alloys with gigapascal yield strength, especially those recently developed body-centered-cubic (bcc) multi-principal-element alloys (MPEAs). The near-zero slope of these curves is often perceived as an apparent lack of strain hardening capability, which would set off necking immediately after yielding due to a violation of the Considère criterion. If so, how could the alloy manage to sustain the large uniform elongation? Here, we resolve this puzzle by re-analyzing the data in terms of true stress versus true strain to demonstrate that the perception above is a misjudgment. Thanks to the MPEA's gigapascal yield strength, the wide plateau does not mean that the strain hardening rate is negligible, but rather is at GPa level, adequate to guarantee no onset of the necking instability. Inside a tensile-strained TiZrNb MPEA, the dislocation density was observed to increase by nearly two orders of magnitude, even after a tensile strain of only 8%, indicating obvious dislocation accumulation as the mechanism for strain hardening to delay necking. All these corroborate that an alloy yielding at GPa stress with a “perfect-plastic” curve is ultra-strong yet highly ductile, despite its very high (i.e., almost ∼1) yield ratio.