The AGILD modeling and inversion have been published in PIERS2005 in Hangzhou. The new 2.5D AGILD and new GL electromagnetic (EM) modeling and inversion are publishing in PIERS2006 in Cambridge of USA. We have proposed the stochastic SGILD modeling and inversion in nano-physics and geophysics using magnetic field equations in 1999 and 2002. In this paper, we present a new stochastic AGILD (SAGILD) EM modeling and inversion and software. We propose a new EM integral equation on the strip domain. Based on the strip integral equation, we propose new stochastic EM field strip integral equations for mean, covariance, and second order mean EM field and parameters. These stochastic EM moments strip integral equations are employed on the boundary strip domain. In the cylindrical and spherical coordinate system, the strip domain includes the boundary and pole strip involving pole � = 0 or r = 0, North Pole � = 0 and South Pole � = �. In the remainder internal domain, we use stochastic moments Galerkin equations. We coupled these equations to construct 3D and 2.5D SAGILD modeling algorithms. By decomposing the variance of the parameter mean, � =� 0 +� 2, � = � 0 +� 2, we derive the new stochastic EM parameters increment moment strip integral equation on the strip sub domain for isotropic or anisotropic random materials. We used the equation pair of EM parameter stochastic integral equations on the strip sub domain and EM parameter stochastic Galerkin equation in remainder sub domain to construct the SAGILD inversion. The SAGILD modeling and inversion have widely applications in the Earth, Sun, and Luna space EM field exploration, nanometer material and material sciences, geophysical and Earthquake exploration, MT, MAIL, VEMP, weather radar imaging, medical MRI and X-ray imaging, and environmental engineering, EM stirring in steel and metal continuous casting, seismic, and finances.