The use of geometric anti-springs (GASs) in MEMS accelerometers can not only boost the sensitivity through the static balancing effect, it also introduces the capability to tune the stiffness post-manufacturing by adjusting the balancing force. This paper addresses the challenge of mechanically fine-tuning the stiffness to create the potential for compensating thermal drift of bias (TDB), whereas previous methods focused on doing so electrostatically. Due to the large effect of TDB on integration errors, creating this new method for compensation could help open up the way to more accurate inertial navigation by high-precision MEMS. As the basis of our research we used an existing accelerometer from Innoseis Sensor Technologies (version G6). Based on a theoretical analysis of the TDB, requirements were formulated and a design was developed through a combination of pseudo-rigid body and finite element modelling. By building a macro-scale prototype, we were able to validate our models. It can therefore be concluded that our design provides 8 additional compression steps ranging between 50 and 54 nm based on the model predictions. This proof-of-concept demonstrates to be a new tool in the design of future accelerometers with GASs.