The diffuse cosmic X-ray background (CXB) is the sum of the emission of discrete sources, mostly massive black-holes accreting matter in active galactic nuclei (AGN). The CXB spectrum differs from the integration of the spectra of individual sources, calling for a large population, undetected so far, of strongly obscured Compton thick AGN. Such objects are predicted by unified models, which attribute most of the AGN diversity to their inclination on the line of sight, and play an important role for the understanding of the growth of black holes in the early Universe. The fraction of obscured AGN at low redshift can be derived from the observed CXB spectrum assuming AGN spectral templates and luminosity functions. We show that high signal-to-noise average hard X-ray spectra, derived from more than a billion seconds of effective exposure time with the Swift/BAT instrument, imply that mildly obscured Compton thin AGN feature a strong reflection and contribute massively to the CXB. A population of Compton thick AGN larger than that effectively detected is not required, as no more than 6\% of the CXB flux can be attributed to them. The stronger reflection observed in mildly obscured AGN suggests that the covering fraction of the gas and dust surrounding their central engines is a key factor in shaping their appearance. These mildly obscured AGN are easier to study at high redshift than Compton thick sources.

Aims. The high energy spectrum of 3C 273 is usually understood in terms of inverse-Compton emission in a relativistic leptonic jet. This model predicts variability patterns and delays that could be tested with simultaneous observations from the radio to the GeV range. Methods. The instruments IBIS, SPI, JEM-X on board INTEGRAL, PCA on board RXTE, and LAT on board Fermi have enough sensitivity to follow the spectral variability of 3C 273 from the keV to the GeV. We looked for correlations between the different energy bands, including radio data at 37 GHz collected at the Mets\"ahovi Radio Observatory and built quasi-simultaneous multiwavelength spectra in the high energy domain when the source is flaring either in the X-rays or in the {\gamma} rays. Results. Both temporal and spectral analysis suggest a two-component model to explain the complete high energy spectrum. X-ray emission is likely dominated by a Seyfert-like component while the {\gamma}-ray emission is dominated by a blazar-like component produced by the relativistic jet. The variability of the blazar-like component is discussed, comparing the spectral parameters in the two different spectral states. Changes of the electron Lorentz factor are found to be the most likely source of the observed variability.

The high energy spectral shape of 3C 273 is usually understood in terms of Inverse-Compton emission in a relativistic leptonic jet. This model predicts variability patterns and delays which could be tested if simultaneous observations are available from the infrared to the GeV range. The instruments IBIS, SPI, JEM-X on board INTEGRAL, PCA on board RXTE and LAT on board Fermi have enough sensitivity to follow the spectral variability from the keV to the GeV and to compare them with model predictions. We are presenting preliminary results on the high energy spectrum of 3C 273 and its variability and compare these results to predictions. We found that a single component is not able to adequately fit the multiwavelength spectrum (5 keV - 10 GeV) when the statistics in Fermi data is sufficient to constraint {\gamma}-ray emission. This suggests that the X-rays emission is not originated in the jet. A possible explanation could be that X-rays are dominated by a Seyfert-like thermal inverse Compton.