The momentum dependence of the superconducting gap in the cuprates has been debated, with most experiments reporting a deviation from a simple $d_{x^2-y^2}$ form in the underdoped regime and a few experiments claiming that a simple $d_{x^2-y^2}$ form persists down to the lowest dopings. We affirm that the superconducting gap function in sufficiently underdoped Bi$_2$Sr$_2$CaCu$_2$O$_{8+delta}$ (Bi-2212) deviates from a simple textit{d}-wave form near the antinode. This is observed in samples where doping is controlled only by oxygen annealing, in contrast to claims that this effect is only seen in cation-substituted samples. Moreover, a quasiparticle peak is present at the antinode down to p$=$0.08, refuting claims that a deviation from a simple textit{d}-wave form is a data analysis artifact stemming from difficulty in assessing a gap in the absence of a quasiparticle.