The phase retrieval hybrid algorithm based on gradient search algorithm and GS algorithm is designed for surface testing of the high dynamic range error beyond one wavelength after the rough polish process. Phase retrieval is a wave front sensing method that uses the intensity distribution to reconstruct the phase distribution of optical field. In the rough polish process of optical surface testing, there is the problem of phase wrapping caused by the high dynamic range error, which makes it hard to be measured by interferometers. In this condition, infrared interferometers are widely used for optical testing, but the operation of infrared interferometer is complicated because of the invisibility of infrared light and it’s very sensitive to the temperature, vibration and turbulence. The phase retrieval hybrid algorithm derives from gradient search algorithm and GS algorithm, aiming to solve the problem of phase wrapping caused by the high dynamic range error. Firstly, phase distribution is described by Zernike polynomials and the coefficients of Zernike polynomials are optimized by the gradient search algorithm to retrieve the low frequency error beyond one wavelength. Then GS algorithm is used to retrieve the high frequency error of small value. In the simulated calculation, the hybrid algorithm is used for surface testing of a spherical mirror with PV 2.42λ , RMS 0.41λ . The retrieved surface is PV 2.53λ , RMS 0.42λ .The simulation result shows that the hybrid algorithm is effective and accurate for optical surface testing of high dynamic range error.