diff --git a/pandas/core/frame.py b/pandas/core/frame.py
index 66399f2b9a5e4..d1be336204593 100644
--- a/pandas/core/frame.py
+++ b/pandas/core/frame.py
@@ -1844,7 +1844,11 @@ def from_records(
                 columns = ensure_index(columns)
             arr_columns = columns
         else:
-            arrays, arr_columns = to_arrays(data, columns, coerce_float=coerce_float)
+            arrays, arr_columns = to_arrays(data, columns)
+            if coerce_float:
+                for i, arr in enumerate(arrays):
+                    if arr.dtype == object:
+                        arrays[i] = lib.maybe_convert_objects(arr, try_float=True)
 
             arr_columns = ensure_index(arr_columns)
             if columns is not None:
diff --git a/pandas/core/internals/construction.py b/pandas/core/internals/construction.py
index bc8e75ceef87b..ec3190d2c86f9 100644
--- a/pandas/core/internals/construction.py
+++ b/pandas/core/internals/construction.py
@@ -500,9 +500,7 @@ def dataclasses_to_dicts(data):
 # Conversion of Inputs to Arrays
 
 
-def to_arrays(
-    data, columns, coerce_float: bool = False, dtype: Optional[DtypeObj] = None
-):
+def to_arrays(data, columns, dtype: Optional[DtypeObj] = None):
     """
     Return list of arrays, columns.
     """
@@ -548,7 +546,7 @@ def to_arrays(
         data = [tuple(x) for x in data]
         content, columns = _list_to_arrays(data, columns)
 
-    content, columns = _finalize_columns_and_data(content, columns, dtype, coerce_float)
+    content, columns = _finalize_columns_and_data(content, columns, dtype)
     return content, columns
 
 
@@ -635,7 +633,6 @@ def _finalize_columns_and_data(
     content: np.ndarray,
     columns: Optional[Union[Index, List]],
     dtype: Optional[DtypeObj],
-    coerce_float: bool,
 ) -> Tuple[List[np.ndarray], Union[Index, List[Axis]]]:
     """
     Ensure we have valid columns, cast object dtypes if possible.
@@ -649,7 +646,7 @@ def _finalize_columns_and_data(
         raise ValueError(err) from err
 
     if len(content) and content[0].dtype == np.object_:
-        content = _convert_object_array(content, dtype=dtype, coerce_float=coerce_float)
+        content = _convert_object_array(content, dtype=dtype)
     return content, columns
 
 
@@ -712,7 +709,7 @@ def _validate_or_indexify_columns(
 
 
 def _convert_object_array(
-    content: List[Scalar], coerce_float: bool = False, dtype: Optional[DtypeObj] = None
+    content: List[Scalar], dtype: Optional[DtypeObj] = None
 ) -> List[Scalar]:
     """
     Internal function ot convert object array.
@@ -720,7 +717,6 @@ def _convert_object_array(
     Parameters
     ----------
     content: list of processed data records
-    coerce_float: bool, to coerce floats or not, default is False
     dtype: np.dtype, default is None
 
     Returns
@@ -730,7 +726,7 @@ def _convert_object_array(
     # provide soft conversion of object dtypes
     def convert(arr):
         if dtype != np.dtype("O"):
-            arr = lib.maybe_convert_objects(arr, try_float=coerce_float)
+            arr = lib.maybe_convert_objects(arr)
             arr = maybe_cast_to_datetime(arr, dtype)
         return arr