If a plot has multi-category axes there is no way to hide a subcategory labels only and the plot doesn't render at all if tickvals and ticktext are used.

Look at the code without multi-categories:

x_names = ['very.long.name.1', 'very.long.name.2', 'very.long.name.3']
fig = go.Figure(go.Bar(
    x=x_names,
    y=[10, 20, 30]))
fig.update_layout(
    xaxis=dict(
        tickvals=x_names,
        ticktext=['', '', ''],
    ))

This works fine and all x-axes labels are hidden. I could achieve the same thing with this:
fig.update_layout(xaxis=dict(showticklabels=False))

However, if this is a multi-category plot it doesn't work at all:

x_names = [['A', 'A', 'B'], ['very.long.name.1', 'very.long.name.2', 'very.long.name.3']]
fig = go.Figure(go.Bar(
    x=x_names,
    y=[10, 20, 30]))
fig.update_layout(
    xaxis=dict(
        tickvals=x_names,
        ticktext=[['A', 'A', 'B'], ['', '', '']],
    ))

Setting showticklabels=False hides everything and I didn't find a way to hide just a single category labels.

What is really surprising this doesn't work either even if this doesn't change anything:

x_names = [['A', 'A', 'B'], ['very.long.name.1', 'very.long.name.2', 'very.long.name.3']]
fig = go.Figure(go.Bar(
    x=x_names,
    y=[10, 20, 30]))
fig.update_layout(
    xaxis=dict(
        tickvals=x_names,
        ticktext=x_names,
    ))

I believe it should work like this:

• tickvals and ticktext work fine with a multi-category plots
• showticklabels should accept a list of booleans, .e.g. [True, False] to be able to hide a given subcategory if this is a multi-category plot