The aim of this work is to use full evolutionary models to derive observational constraints on the mass loss rate of the upper Asymptotic Giant Branch ( AGB ) stars .
The observations used to constrain the models are : i ) the relative number of luminous Lithium rich AGBs in the Magellanic Clouds , with respect to the total number of AGBs populating the luminosity range -6 \geq M _ { bol } \geq - 7 ; ii ) the s–process enhancement of the same sample .
The calibration of the mass loss rate we obtain gives feedbacks on the interpretation of observational data of obscured AGBs , and allows us to provide consistent lithium yields for these stars , to be used to constrain the galactic chemical evolution .

We find that : a ) we can put lower and upper limits to the mass loss rate during the AGB phase ; b ) after a “ visible ” phase , the models evolve into a phase of strong mass loss , which can be identified with the obscured OH/IR stars accessible only in the infrared ; the models well reproduce the Period– M _ { bol } loci of the obscured AGBs ( Wood et al .
1992 ) ; c ) the most massive AGBs ( mass of progenitors , hereinafter M _ { ZAMS } , \sim 6 M _ { \odot } ) are extremely luminous ( M _ { bol } \sim - 7.2 to -7.5 ) ; d ) The lithium yield increases with the mass loss rate and with the total stellar mass , being maximum for AGB stars close to the lower limit for carbon semi-degenerate ignition .
However , the mass loss calibration obtained in this work implies that massive AGBs do not contribute significantly to the lithium enrichment of the interstellar medium .