diff --git a/hpc/batch_gpu/submit_imaging_mge_a100_hst_fp64 b/hpc/batch_gpu/submit_imaging_mge_a100_hst_fp64
index e827571..5f0f3ee 100755
--- a/hpc/batch_gpu/submit_imaging_mge_a100_hst_fp64
+++ b/hpc/batch_gpu/submit_imaging_mge_a100_hst_fp64
@@ -30,10 +30,6 @@ export XLA_PYTHON_CLIENT_PREALLOCATE=false
 export JAX_ENABLE_X64=True
 export NUMBA_CACHE_DIR=/tmp/numba_cache
 export MPLCONFIGDIR=/tmp/matplotlib
-# Skip latent-variable computation: autolens_profiling/config/latent.yaml
-# enables total_lens_flux_mujy which needs `magzero=` on the Analysis.
-# Latents aren't part of the search profile, so we bypass entirely.
-export PYAUTO_SKIP_LATENTS=1
 
 nvidia-smi