IASI and CrIS, hyperspectral infrared instruments on board polar-orbiting operational meteorological satellites, have been assimilated in Numerical Weather Prediction models for more than a decade, providing huge improvements in forecast skills. In the same time, level 2 products from these sensors have been introduced in some atmospheric composition numerical prediction models, for carbon monoxide, for instance. Previous work has led to a first set of evaluation of IASI and CrIS radiances assimilation in the chemistry transport model MOCAGE for ozone, showing a good impact but also highlighting the need of further improvements in assimilation settings. The present work will introduce recent steps towards a better use of infrared radiances in MOCAGE for ozone and also for carbon monoxide. First the evaluation of new background error statistics will be presented, using the so-called NMC method both applied to ozone and carbon monoxide. Then the channel selection will be presented for both instruments and both species. Assimilation trials in MOCAGE will be evaluated against independent observations such as ozone-sondes, IAGOS measurements, MLS profiles ans ground-based in situ observations.