Tony has things correctly stated. I thought I'd just add another way of writing similar things without getting mired. Let's redraw the circuits:
simulate this circuit – Schematic created using CircuitLab
I've drawn the regular analog meter movement on the top, and its equivalent. Here, you can see that the meter does actually inject a potentially significant resistance in series (which may upset the measurement, depending on circumstances.) The analog meter movement here presents \$500\:\Omega\$ of DC resistance.
I've drawn the opamp circuit on the bottom, and its equivalent. Here, you can see that the effective series resistance may be very significantly diminished, now. (\$A_O\$ is the open loop gain of the opamp.) The open loop gain of the opamp would have to be as bad as 100 to present the same series resistance. But that's not the usual case. Open loop gains are much, much higher in even the cheapest opamps. (I've neglected offset voltage and current and bias currents, to focus on the central point.)
By the way, you can often use much cheaper analog panel meters as a result, as well, and still get significant reductions in the impact on a circuit under measurement using an opamp. And opamps, even expensive ones, are often a lot cheaper than a very high quality analog D'Arsonval movement will be. I suspect this has largely killed wide-scale production of sensitive D'Arsonval movements, despite the somewhat more recent advent of very powerful Neodymium permanent magnets.