We report the discovery of two new transiting hot Jupiters, KOI-135b and KOI-204b, that were previously identified as planetary candidates by Borucki et al. 2011, and, independently of the Kepler team, confirm the planetary nature of Kepler-17b, recently announced by Desert et al. 2011. Radial-velocity measurements, taken with the SOPHIE spectrograph at the OHP, and Kepler photometry (Q1 and Q2 data) were used to derive the orbital, stellar and planetary parameters. KOI-135b and KOI-204b orbit their parent stars in 3.02 and 3.25 days, respectively. They have approximately the same radius, Rp=1.20+/-0.06 R_jup and 1.24+/-0.07 R_jup, but different masses Mp=3.23+/-0.19 M_jup and 1.02+/-0.07 M_jup. As a consequence, their bulk densities differ by a factor of four, rho_p=2.33+/-0.36 g.cm^-3 (KOI-135b) and 0.65+/-0.12 g.cm-3 (KOI-204b). Our SOPHIE spectra of Kepler-17b, used both to measure the radial-velocity variations and determine the atmospheric parameters of the host star, allow us to refine the characterisation of the planetary system. In particular we found the radial-velocity semi-amplitude and the stellar mass to be respectively slightly smaller and larger than Desert et al. These two quantities, however, compensate and lead to a planetary mass fully consistent with Desert et al.: our analysis gives Mp=2.47+/-0.10 M_jup and Rp=1.33+/-0.04 R_jup. We found evidence for a younger age of this planetary system, t<1.8 Gyr, which is supported by both evolutionary tracks and gyrochronology. Finally, we confirm the detection of the optical secondary eclipse and found also the brightness phase variation with the Q1 and Q2 Kepler data. The latter indicates a low redistribution of stellar heat to the night side (<16% at 1-sigma), if the optical planetary occultation comes entirely from thermal flux. The geometric albedo is A_g<0.12 (1-sigma).