To explore sensitivity to initial conditions we perturb those conditions and watch for departure from the original calculation. First we give the top particle with mass 3 a velocity to the right of 10–2. At time t = 32 the images are just a few pixels off but by t=34 the dance has changed and by t=35 masses 3 and 4 have eloped:
Starting at t=32:

Starting mass 3 with vx = 10–3 the difference becomes visible at about t = 38. Here are the images starting at t = 32.
I don’t know if mass 3 escapes.

For mass 3 starting with vx = 10–4 we trace until t = 90. Starting at about t = 40 differences begin to be visible. The following is for 40 < t < 90.

For mass 3 starting with vx = 10–5 we trace until t = 88. The difference appears at t = 46. The following is for 40 < t < 88.

For mass 3 starting with vx = 10–6 we trace until t = 88. The difference appears at about t = 52. The real divergence follows a near expulsion and the final encounter is much changed. Starting from t = 48:

For mass 3 starting with vx = 10–7 we trace until t = 88. The difference appears at about t = 59.
Starting from t = 56: