It’s hard to help you, because there is no universal 3D Gibbs free energy landscape.
To generate such a landscape, you would have to say what want to be looking at, e.g., a distance, an angle, or projections onto principal components.
Once you determined what the essential observables of your system are, you would calculate them frame-by-frame from your simulation – imagine a table like this:
frame
distance C_1 - N_2
another distance
projection on principal compent 1
1
0.22
0.13
0.54
…
…
…
…
10 000
0.21
0.03
-0.9
Then, you can use some type of density estimation tool to plot the three-dimensional density. For this, I would rather suggest some combination of python tools (especially matplotlib, pandas and seaborn are helpful here) than the built in GROMACS tools.