As mentioned by heropup, Borwein and Zucker found closed forms for
$\Gamma(n/24)$, for integer $n$, using the complete elliptic integral of the first kind; see Wikipedia for details, with references.
Elliptic integrals can be rapidly evaluated to high precision using algorithms based on the AGM, the arithmetic-geometric mean, which converges quadratically.
The AGM is defined as follows. Let $x_0=x, y_0=y$. Then,
$$\begin{align}
x_{i+1} & = (x_i + y_i) / 2 \\
y_{i+1} & = \sqrt{x_i y_i}
\end{align}$$
These sequences rapidly converge on the same value:
$$\operatorname{AGM}(x, y) = \lim_{i\to \infty} \, x_i = \lim_{i\to \infty} \, y_i$$
Elliptic integral notation can be confusing. These integrals are sometimes written using the modulus $m$, and sometimes using the parameter $k$, where $m = k^2$.
The complete elliptic integral of the first kind can be calculated as
$$K(m) = \frac{\pi/2}{\operatorname{AGM}(1, \sqrt{1-m})}$$
Wikipedia gives a couple of equivalent expressions for $\Gamma(1/3)$ in terms of $K()$. Here's a breakdown into a few logical steps.
$$\begin{align}
m & = \frac{2 - \sqrt{3}}{4} \\
q & = \frac{\pi/2}{\operatorname{AGM}(1, \sqrt{1-m})} \\
g & = \frac{q \pi \sqrt[3]{128}}{\sqrt[4]{3}} \\
\Gamma\left(\frac{1}{3}\right) & = \sqrt[3]{g}
\end{align}$$
As mentioned, the AGM converges quadratically. i.e., the number of correct digits doubles on each loop of the iteration. For the $\Gamma(1/3)$ calculation, two loops are adequate for the precision of a typical pocket scientific calculation, three loops are adequate for standard double-precision floating-point numbers.
Here's a simple implementation of the above algorithm in Python.
Code
""" gamma(1/3), using AGM
Written by PM 2Ring 2024.06.10
Pure Python
"""
from itertools import count
from math import sqrt, pi
def agm(g, a, eps=1e-9):
for i in count(1):
g, a = sqrt(g * a), (g + a) / 2
print(i, g, a)
if a - g < eps:
break
return (a + g) / 2
m = (2 - sqrt(3)) / 4
print(f"{m=}")
a = agm(sqrt(1 - m), 1)
print(f"{a=}")
q = pi/2 / a
print(f"{q=}")
g3 = pi * q * (128**(1/3) / 3**(1/4))
g = g3**(1/3)
print(g)
Output
m=0.0669872981077807
1 0.9828152554214186 0.9829629131445341
2 0.9828890815101808 0.9828890842829763
3 0.9828890828965786 0.9828890828965786
a=0.9828890828965786
q=1.59814200211254
2.6789385347077475
Here's a live version of that script, running on the SageMathCell server. And here's a Sage version, using arbitrary precision arithmetic. Of course, Sage also provides the gamma function, the AGM, and a full complement of elliptic integrals and functions.
Here's $\Gamma(1/3)$ using the Sage code, to 300 bits precision.
2.
67893 85347 07747 63365 56929 40974 67764 41286 89377 95730 11009 50428 32759 04176 10167 74381 95409 8289
Incidentally, it's possible to calculate arbitrary values of the gamma function using mostly elementary arithmetic (apart from a couple of exponentiations), via the continued fractions of the upper and lower incomplete gamma functions. However, that algorithm doesn't converge quickly, so it's a bit too tedious for manual calculation. I have details & code in this answer.