The Bordeaux group on WP-811-20000 has now 5 poeple for 1.2 FTE. GBOG observations still on-going on NARVAL. Analyses of spectra with SME code started. Participation to LUMBA in Gaia-ESO survey will help to define the good inputs for SME. RAVE and SEGUE have already faced the problem of calibrating APs. Several other surveys coming soon with same issue. RAVE 3rd release (Siebert et al. 2011) provides Teff, logg, [M/H], [α/Fe] for ∼40000 stars. Internal stability assessed with repeat observations. Compari- son to external data with catalogue of 102 reference stars included in RAVE observations + 227 RAVE targets observed at high resolution. Mean offset and dispersion -63, 250 K in Teff, -0.1, 0.43 dex in logg. [M/H] calibrated as a function of derived metallicity, alpha enhancement, Teff, logg, S/N, large dispersion, not able to retreive high metallicties. SEGUE provides APs for 240000 faint stars. 4 papers on SSPP and its validation. Uses several different methods then decision tree to adopt final AP. Teff, log g, [Fe/H] at precisions of 141 K, 0.23 dex, 0.23 dex. Comparison to 125 SEGUE stars observed at high resolution gives external errors of 130 K, 0.21 dex, and 0.11 dex. Mean metallicity of 5 GCs + 3 OCs compared to literature values. SSPP improving with new external data, now 343 SEGUE targets observed at high S/N with aim of 500 but not analysed homogeneously. Residuals AP(SSPP)-AP(HR) show strong sys- tematics, specially in [Fe/H]. (This may be an issues with either the AP estimation of the high resolution spectra - as suggested in the paper - but also in the SSPP algorithms.) Preliminary results of AMBRE on FEROS (R=48000) provided APs for 23000 stars. Comparison to liter- ature is good but essentially limited to solar metallicity dwarfs (Worley et al. 2010). Offsets and dispersion also visible in literature values. Gaia-ESO survey plans to allocate 10 nights to calibrating clusters. Standard fields being defined. Gaia AP reference stars to be included in pointings when possible. Calibration of CU8 algorithm should use as many AP reference stars as possible. All previous surveys and calibration samples will be included in Gaia. Efforts done by other surveys useful for us. A schedule has to be defined for producing the reference grid. Ulrike Heiter and Caroline Soubiran will inform CU8 how many stars could be analysed and used in the first run of Ap- sis. For this run, the APs of the reference stars don't need to be of very high accuracy and homogenity, as they will be improved prior to use in producing the final Gaia catalogue.< Réduire