Is there an issue with the AD8610 spice model?
When I compare the ADA4610, AD8625 and AD8610, the output (Vout3) of the 8610 is heavily dependent on the input resistor R14.
For 1K or higher the output goes to the -rail.
This occurs when using +- 10V rails (as the design will be implemented)
If i change the rails in the model to +-13V the problem does not occur.
I rechecked the AD8610 spice model with the same schematic and simulation as your image above.
I did not see the railing you describe in AD8610 when R14 is 1k or higher and supply is +/-10.
I'm attaching my LTspice files for your reference.AD8610_spicemodel_issue.zip