Geophysical parameters from the IASI instrument on Metop-A are essential products provided from EUMETSAT's Central Facility in near real time. They include vertical profiles of temperature and humidity, related cloud information, surface emissivity and temperature, and atmospheric composition parameters (CO, ozone and several other trace gases). The IASI L2 processor version 5 has been running routinely since September 2010. We give here a short summary of the most recent related validation studies, including results from the Metop-B/IASI L2 Cal/Val activities which started end January 2013, after the conclusion of the first phase of the L1c commissioning. We then introduce the current developments and the future evolutions of the IASI L2 processor operated at EUMETSAT towards the version 6 (v6), which release is scheduled second half of 2013. In particular we will address the integration of new atmospheric composition products from ULB/LATMOS in the scope of O3M-SAF CDOP-2, with first the retrieval of CO profiles. Only the CO total column has been retrieved until version 5. The cloud screening has been improved with the addition of a third detection algorithm based on artificial neural networks and independent from NWP forecasts. We will discuss here the use of concurrent cloud detection methods and the relationship between the cloudiness estimate and the sensing products yield and quality. A quality indicator is associated to the retrieved geophysical parameters in v6. Lastly, new retrieval methods for temperature and water-vapour sounding have been developed, including joint use of micro-wave (MW) and infrared (IR) measurements. The statistical MW+IR retrieval scheme, developed for v6, takes advantage of AMSU and MHS measurements collocated with IASI L1c spectra. It allows for sounding below the clouds in most cloud contaminated IFOVs with improved performances (typically about 1K rms) as compared to the current operational IR-based partly-cloudy retrieval. The MWIR-derived temperature and humidity profiles are used to constrain the retrieval of temperature, humidity and ozone 1D-Var retrievals in clear sky. The latter are performed with reconstructed radiances from a new channel selection and a scene-dependent bias correction. Initial results with the version 6 using ECMWF analyses as reference have shown that the temperature profiles are improved by 0.2 to 0.5K rms in the lower troposphere and that the precision in water-vapour profiles is increased by 30 to 50% in the boundary layer.