Try solve a system of equation like this one.
\begin{cases} (O_x -A_x)^2+(O_y-A_y)^2+(O_z-Az)^2=x^2 \\ (O_x -B_x)^2+(O_y-B_y)^2+(O_z-Bz)^2=y^2 \\ (O_x -C_x)^2+(O_y-C_y)^2+(O_z-Cz)^2=z^2 \end{cases}
Is there any way to express $O_x, O_y, O_z$ in terms of $A_x, A_y, A_z, B_x, B_y, B_z,C_x,C_y,C_z, x, y, z$?
Here is a geometric solution. I would use a numerical approach, since writing out a single formula would be ugly.
$A,B,C$ are points that define a plane, and $O$ is a point outside this plane. First, make sure that the corresponding sides obey triangle inequality. I am going to skip this step. The lengths of the sides of the tetrahedron are $a=|BC|$, $b=|AC|$, $c=|AB|$, $x=|OA|$, $y=|OB|$, and $z=|OC|$. All these are known.
Then let's calculate the angles between any pair of sides, using generalized Pythagoras' theorem. For example $$\cos\angle OAB=\frac{x^2+c^2-y^2}{2xc}$$ Similarly, $$\cos\angle CAB=\frac{b^2+c^2-a^2}{2bc}$$ Now let's say $P$ is the point on plane $ABC$ where $OP$ perpendicular to $ABC$,$M$ is the point on $AB$ where a plane containing $OPM$ is perpendicular to $AB$. Then $$|AM|=x\cos\angle OAB=\frac{x^2+c^2-y^2}{2c}$$ This allows you to calculate the coordinates of point $M$: $$M_i=\frac{B_i-A_i}{c}|AM|+A_i$$ Now we calculate the lengths of $OM$ and $PM$. $$|OM|=x\sin\angle OAB=x\sqrt{1-\cos^2\angle OAB}\\|PM|=*********************$$
At this point, you have all lengths in the $\triangle POM$, which allows you to calculate any angles. In particular $$\cos\angle PMO=\frac{|PM|^2+|OM|^2-|OP|^2}{2|PM||OM|}$$
Now let's calculate point $O$. For this, we use the vector product. We know that the $COM$ plane is perpendicular to $AB$ so $\vec{OM}\times\vec{CM}$ is along vector $\vec{AB}$. We also know the magnitude of the cross product, so we write $$\vec{OM}\times\vec{PM}=\pm|OM||PM|\sin\angle PMO\frac{\vec{AB}}{c}$$
All you need is to write the components of the vector product and you have two linear systems in $O_{x,y,z}$, one for each of the signs in the above equation. Writing out explicitly the formula for this terms is tedious, but it's trivial to solve numerically.