As part of the Global Earthquake Model (GEM) effort to improve global seismic hazard models, we present a new global geodetic strain rate model. This model (GSRM v. 2) is a vast improvement on the previous model from 2004 (v. 1.2). The model is still based on a finite-element type approach and has deforming cells in between the assumed rigid plates. The new model contains ~144,700 cells of 0.25° by 0.2° dimension. We redefined the geometries of the deforming zones based on the definitions of Bird (2003) and Chamot-Rooke and Rabaute (2006). We made some adjustments to the grid geometry at places where seismicity and/or GPS velocities suggested either the presence of deforming areas or a rigid block where those previous studies did not. GSRM v.2 includes 50 plates and blocks, including many not considered by Bird (2003). The new GSRM model is based on over 20,700 horizontal geodetic velocities at over 17,000 unique locations. The GPS velocity field consists of a 1) Over 6500 velocities derived by the University of Nevada, Reno, for CGPS stations for which >2.5 years of RINEX data are available until April 2013, 2) ~1200 velocities for China from a new analysis of all data from the Crustal Movement Network of China (CMONOC), and 3) about 13,000 velocities from 212 studies published in the literature or made otherwise available to us. Velocities from all studies were combined into the same reference frame by a 6-parameter transformation using velocities at collocated stations. We model co-seismic jumps while estimating velocities, ignore periods of post-seismic deformation, and exclude time-series that reflect magmatic and anthropogenic activity. GPS velocities were used to estimate angular velocities for 36 of the 50 rigid plates and blocks (the rest being taken from the literature), and these were used as boundary conditions in the strain rate calculations. For the strain rate calculations we used the method of Haines and Holt. In order to fit the data equally well in slowly and rapidly deforming areas, we first calculated a very smooth model by setting the a priori variances of the strain rate components very low. We then used this model as a proxy for the a priori standard deviations of the final model, at least for the areas that are well constrained by the GPS data. We will show examples of the strain rate and velocity field results. We will also highlight how and where the results can be viewed and accessed through a dedicated webportal (gsrm2.unavco.org). New GPS velocities (in any reference frame) can be uploaded to a new tool and displayed together with velocities used in GSRM v.2 in 53 reference frames (http://facility.unavco.org/data/maps/GPSVelocityViewer/GSRMViewer.html) .