Hello, I am designing this Bandgap topology, but I have a problem, although the currents and voltages of the current mirror coincide, and the vref is 1.1v, I have a rather strange curve, since it should look like a parabola, does anyone know what can to be?

It looks like you have too much PTAT added to the Vbe. Try to lower the PTAT content by tweaking R1 and/or R2 until you have a the first-order term cancelled, and only 2nd order.

but if I reduce the resistances it drops or increases vref, in addition to changing the currents that must work at a certain design specification

This startup circuit seems strange. Have you tried to run VDD sweep simulation to see the reference voltage variation?

Yes, tweak until you kind of an equripple response. Now you optimized for models, usually a re-tuning after measurements is needed to tweak for actual performance. You will be a bit off, not much I guess.

Could you explain to me the actual performance? How do I know what I should have done?

Make the same simulations with three VDD points: 0.9, 1.0 and 1.1 typical VDD. How does its output behave?