Oh I see, I solved for the angular frequency, not the angular frequency of small oscillations.
This is what I should have done:
Lag = T - U =1/2 I * w^2 - m*g*L Cos (theta) D/Dt = full derivative
D/Dt (dLag/dw) - dLag/dtheata = 0 d/dx = partial derivative
Which gives I*a - m*g*L Sin(theta) = 0
From that point on it's just like your solution.