diff --git a/scripts/imaging/visualization/modeling_visualization_jit.py b/scripts/imaging/visualization/modeling_visualization_jit.py
index f509d37c..917bcded 100644
--- a/scripts/imaging/visualization/modeling_visualization_jit.py
+++ b/scripts/imaging/visualization/modeling_visualization_jit.py
@@ -312,19 +312,21 @@
 result = search.fit(model=model_mge2, analysis=analysis_mge2)
 
 # The Nautilus output goes to output/<path_prefix>/<name>/<hash>/image/
-# The lens quick-update visualizer writes fit_quick.png (via subplot_fit_quick)
-# to that image folder during each quick update.
+# Quick updates write the normal fit subplot (fit.png) to that image folder —
+# the same file full visualization writes. Quick updates run on the main
+# thread by default, so the KeyError regression this script guards would
+# crash the search rather than be swallowed.
 output_search_root = with_test_mode_segment(Path("output")) / output_root / "mge_linear"
-produced_pngs = list(output_search_root.rglob("fit_quick.png"))
-print(f"fit_quick.png files produced: {len(produced_pngs)}")
+produced_pngs = list(output_search_root.rglob("fit.png"))
+print(f"fit.png files produced: {len(produced_pngs)}")
 for p in produced_pngs:
     print(f"  {p}")
 assert len(produced_pngs) > 0, (
-    f"no fit_quick.png produced under {output_search_root} — "
-    "quick-update visualization did not fire"
+    f"no fit.png produced under {output_search_root} — "
+    "fit visualization did not fire"
 )
 print(
     "\nPASS: jit-cached fit_for_visualization fires during Nautilus quick updates "
-    "with MGE linear profiles, fit_quick.png written, no KeyError from "
+    "with MGE linear profiles, fit.png written, no KeyError from "
     "linear_light_profile_intensity_dict lookup."
 )
diff --git a/scripts/interferometer/visualization/modeling_visualization_jit.py b/scripts/interferometer/visualization/modeling_visualization_jit.py
index 7a85e70c..ac473837 100644
--- a/scripts/interferometer/visualization/modeling_visualization_jit.py
+++ b/scripts/interferometer/visualization/modeling_visualization_jit.py
@@ -321,17 +321,19 @@
 result = search.fit(model=model_mge2, analysis=analysis_mge2)
 
 # The Nautilus output goes to output/<path_prefix>/<name>/<hash>/image/
-# The lens quick-update visualizer writes fit_quick.png during each quick update.
-produced_pngs = list(output_search_root.rglob("fit_quick.png"))
-print(f"fit_quick.png files produced: {len(produced_pngs)}")
+# Quick updates write the normal fit subplot (fit.png) — the same file full
+# visualization writes. Quick updates run on the main thread by default, so a
+# regression in the quick path would crash the search rather than be swallowed.
+produced_pngs = list(output_search_root.rglob("fit.png"))
+print(f"fit.png files produced: {len(produced_pngs)}")
 for p in produced_pngs:
     print(f"  {p}")
 assert len(produced_pngs) > 0, (
-    f"no fit_quick.png produced under {output_search_root} — "
-    "quick-update visualization did not fire"
+    f"no fit.png produced under {output_search_root} — "
+    "fit visualization did not fire"
 )
 print(
     "\nPASS: jit-cached fit_for_visualization fires during Nautilus quick updates "
-    "with MGE linear profiles, fit_quick.png written, no KeyError from "
+    "with MGE linear profiles, fit.png written, no KeyError from "
     "linear_light_profile_intensity_dict lookup."
 )
diff --git a/scripts/point_source/visualization/modeling_visualization_jit.py b/scripts/point_source/visualization/modeling_visualization_jit.py
index 68e79ebd..ac035675 100644
--- a/scripts/point_source/visualization/modeling_visualization_jit.py
+++ b/scripts/point_source/visualization/modeling_visualization_jit.py
@@ -276,16 +276,18 @@
 result = search.fit(model=model2, analysis=analysis_run)
 
 # Nautilus writes quick-update images to output/<path_prefix>/<name>/<hash>/image/
-# The lens quick-update visualizer writes fit_quick.png (via subplot_fit_quick).
-produced_pngs = list(output_search_root.rglob("fit_quick.png"))
-print(f"fit_quick.png files produced: {len(produced_pngs)}")
+# Quick updates write the normal fit subplot (fit.png) — the same file full
+# visualization writes. Quick updates run on the main thread by default, so a
+# regression in the quick path would crash the search rather than be swallowed.
+produced_pngs = list(output_search_root.rglob("fit.png"))
+print(f"fit.png files produced: {len(produced_pngs)}")
 for p in produced_pngs:
     print(f"  {p}")
 assert len(produced_pngs) > 0, (
-    f"no fit_quick.png produced under {output_search_root} — "
-    "quick-update visualization did not fire"
+    f"no fit.png produced under {output_search_root} — "
+    "fit visualization did not fire"
 )
 print(
     "\nPASS: jit-cached fit_for_visualization fires during Nautilus quick updates "
-    f"for point source, fit_quick.png written."
+    f"for point source, fit.png written."
 )