We present the analysis of microlensing event OGLE-2006-BLG-284, which has a lens system that consists of two stars and a gas giant planet with a mass ratio of $q_p = (1.26pm 0.19) times 10^{-3}$ to the primary. The mass ratio of the two stars is $q_s = 0.289pm 0.011$, and their projected separation is $s_s = 2.1pm 0.7,$AU, while the projected separation of the planet from the primary is $s_p = 2.2pm 0.8,$AU. For this lens system to have stable orbits, the three-dimensional separation of either the primary and secondary stars or the planet and primary star must be much larger than that these projected separations. Since we do not know which is the case, the system could include either a circumbinary or a circumstellar planet. Because there is no measurement of the microlensing parallax effect or lens system brightness, we can only make a rough Bayesian estimate of the lens system masses and brightness. We find host star and planet masses of $M_{L1} = 0.35^{+0.30}_{-0.20},M_odot$, $M_{L2} = 0.10^{+0.09}_{-0.06},M_odot$, and $m_p = 144^{+126}_{-82},M_oplus$, and the $K$-band magnitude of the combined brightness of the host stars is $K_L = 19.7^{+0.7}_{-1.0}$. The separation between the lens and source system will be $sim 90,$mas in mid-2020, so it should be possible to detect the host system with follow-up adaptive optics or Hubble Space Telescope observations.