diff --git a/beets/dbcore/db.py b/beets/dbcore/db.py
index b1dc981b2..2b32a2076 100755
--- a/beets/dbcore/db.py
+++ b/beets/dbcore/db.py
@@ -35,8 +35,9 @@ from typing_extensions import (
 from unidecode import unidecode
 
 import beets
+from beets.util.functemplate import get_template
 
-from ..util import cached_classproperty, functemplate
+from ..util import cached_classproperty
 from . import types
 from .query import MatchQuery, TrueQuery
 from .sort import NullSort
@@ -688,20 +689,13 @@ class Model(ABC, Generic[D]):
         """
         return self._formatter(self, included_keys, for_path)
 
-    def evaluate_template(
-        self, template: str | functemplate.Template, for_path: bool = False
-    ) -> str:
-        """Evaluate a template (a string or a `Template` object) using
-        the object's fields. If `for_path` is true, then no new path
-        separators will be added to the template.
+    def evaluate_template(self, fmt: str, for_path: bool = False) -> str:
+        """Evaluate a format string using the object's fields.
+
+        If `for_path` is true, then no new path separators are added to the template.
         """
         # Perform substitution.
-        if isinstance(template, str):
-            t = functemplate.template(template)
-        else:
-            # Help out mypy
-            t = template
-        return t.substitute(
+        return get_template(fmt).substitute(
             self.formatted(for_path=for_path), self._template_funcs()
         )
 
diff --git a/beets/library/models.py b/beets/library/models.py
index 7d8172014..95b623833 100644
--- a/beets/library/models.py
+++ b/beets/library/models.py
@@ -26,7 +26,6 @@ from beets.util import (
     syspath,
 )
 from beets.util.deprecation import maybe_replace_legacy_field
-from beets.util.functemplate import Template, template
 from beets.util.pathformats import PF_KEY_DEFAULT
 
 from .exceptions import FileOperationError, ReadError, WriteError
@@ -102,10 +101,9 @@ class LibModel(dbcore.Model["Library"]):
         super().add(lib)
 
     def __format__(self, spec: str) -> str:
-        if not spec:
-            spec = beets.config[self._format_config_key].as_str()
-        assert isinstance(spec, str)
-        return self.evaluate_template(spec)
+        return self.evaluate_template(
+            spec or beets.config[self._format_config_key].as_str()
+        )
 
     def __str__(self) -> str:
         return format(self)
@@ -546,8 +544,8 @@ class Album(LibModel):
         image = bytestring_path(image)
         item_dir = item_dir or self.item_dir()
 
-        filename_tmpl = template(beets.config["art_filename"].as_str())
-        subpath = self.evaluate_template(filename_tmpl, True)
+        filename_tmpl = beets.config["art_filename"].as_str()
+        subpath = self.evaluate_template(filename_tmpl, for_path=True)
         if beets.config["asciify_paths"]:
             subpath = util.asciify_path(subpath)
         subpath = util.sanitize_path(subpath, replacements=self.db.replacements)
@@ -1250,13 +1248,8 @@ class Item(LibModel):
                     break
             else:
                 assert False, "no default path format"
-        if isinstance(path_format, Template):
-            subpath_tmpl = path_format
-        else:
-            subpath_tmpl = template(path_format)
-
         # Evaluate the selected template.
-        subpath = self.evaluate_template(subpath_tmpl, True)
+        subpath = self.evaluate_template(path_format, for_path=True)
 
         if beets.config["asciify_paths"]:
             subpath = util.asciify_path(subpath)
diff --git a/beets/test/helper.py b/beets/test/helper.py
index ffd623eb8..9f72bad00 100644
--- a/beets/test/helper.py
+++ b/beets/test/helper.py
@@ -39,7 +39,6 @@ from beets.library import Item, Library
 from beets.test import _common
 from beets.ui.commands.import_.session import TerminalImportSession
 from beets.util import MoveOperation, clean_module_tempdir, syspath
-from beets.util.functemplate import template
 
 if TYPE_CHECKING:
     from collections.abc import Iterable, Iterator, Sequence
@@ -614,12 +613,9 @@ class ImportHelper(TestHelper, ImporterMixin):
         super().setup_beets()
         self.import_media = []
         self.lib.path_formats = [
-            ("default", template(os.path.join("$artist", "$album", "$title"))),
-            ("singleton:true", template(os.path.join("singletons", "$title"))),
-            (
-                "comp:true",
-                template(os.path.join("compilations", "$album", "$title")),
-            ),
+            ("default", os.path.join("$artist", "$album", "$title")),
+            ("singleton:true", os.path.join("singletons", "$title")),
+            ("comp:true", os.path.join("compilations", "$album", "$title")),
         ]
 
 
diff --git a/beets/ui/commands/modify.py b/beets/ui/commands/modify.py
index 18ea099f5..e03b2de2f 100644
--- a/beets/ui/commands/modify.py
+++ b/beets/ui/commands/modify.py
@@ -7,7 +7,6 @@ from typing import TYPE_CHECKING, NamedTuple
 from beets import library, ui
 from beets.dbcore import types
 from beets.exceptions import UserError
-from beets.util import functemplate
 from beets.util.deprecation import maybe_replace_legacy_field
 
 from .utils import do_query
@@ -69,16 +68,15 @@ def modify_items(lib, mods, dels, query, write, move, album, confirm, inherit):
     # objects.
     ui.print_(f"Modifying {len(objs)} {'album' if album else 'item'}s.")
     changed = []
-    templates = {}
-    for key, mod in mods.items():
-        templates[key] = functemplate.template(mod.value)
     for obj in objs:
-        obj_mods = {}
-        for key, mod in mods.items():
-            parsed_value = model_cls._parse(
-                key, obj.evaluate_template(templates[key])
+        obj_mods = {
+            key: mod.apply(
+                obj,
+                key,
+                model_cls._parse(key, obj.evaluate_template(mod.value)),
             )
-            obj_mods[key] = mod.apply(obj, key, parsed_value)
+            for key, mod in mods.items()
+        }
         if print_and_modify(obj, obj_mods, dels) and obj not in changed:
             changed.append(obj)
 
diff --git a/beets/util/functemplate.py b/beets/util/functemplate.py
index 5547bcb0e..e7451fb3b 100644
--- a/beets/util/functemplate.py
+++ b/beets/util/functemplate.py
@@ -16,9 +16,9 @@ from __future__ import annotations
 
 import ast
 import dis
-import functools
 import re
 import types
+from functools import lru_cache
 
 SYMBOL_DELIM = "$"
 FUNC_DELIM = "%"
@@ -496,8 +496,8 @@ def _parse(template):
     return Expression(parts)
 
 
-@functools.lru_cache(maxsize=128)
-def template(fmt) -> Template:
+@lru_cache(maxsize=128)
+def get_template(fmt: str) -> Template:
     return Template(fmt)
 
 
diff --git a/beets/util/pathformats.py b/beets/util/pathformats.py
index 275f0f2cb..7d1a6faf5 100644
--- a/beets/util/pathformats.py
+++ b/beets/util/pathformats.py
@@ -2,14 +2,10 @@ from __future__ import annotations
 
 from typing import TYPE_CHECKING
 
-from .functemplate import template
-
 if TYPE_CHECKING:
     import confuse
 
-    from .functemplate import Template
-
-    PathFormat = tuple[str, Template]
+    PathFormat = tuple[str, str]
 
 
 # Special path format key.
@@ -25,7 +21,4 @@ def get_path_formats(subview: confuse.Subview) -> list[PathFormat]:
     part of ``paths``. This keeps inherited defaults such as ``default``,
     ``comp``, and ``singleton`` available unless they are explicitly replaced.
     """
-    return [
-        (PF_KEY_QUERIES.get(q, q), template(v.as_str()))
-        for q, v in subview.items()
-    ]
+    return [(PF_KEY_QUERIES.get(q, q), v.as_str()) for q, v in subview.items()]
diff --git a/beetsplug/bench.py b/beetsplug/bench.py
index d1f71c7ad..3beca7928 100644
--- a/beetsplug/bench.py
+++ b/beetsplug/bench.py
@@ -6,7 +6,6 @@ import timeit
 from beets import importer, plugins, ui
 from beets.autotag import tag_album
 from beets.plugins import BeetsPlugin
-from beets.util.functemplate import Template
 from beets.util.pathformats import PF_KEY_DEFAULT
 from beetsplug._utils import vfs
 
@@ -17,10 +16,7 @@ def aunique_benchmark(lib, prof):
 
     # Measure path generation performance with %aunique{} included.
     lib.path_formats = [
-        (
-            PF_KEY_DEFAULT,
-            Template("$albumartist/$album%aunique{}/$track $title"),
-        )
+        (PF_KEY_DEFAULT, "$albumartist/$album%aunique{}/$track $title")
     ]
     if prof:
         cProfile.runctx(
@@ -35,7 +31,7 @@ def aunique_benchmark(lib, prof):
 
     # And with %aunique replaced with a "cheap" no-op function.
     lib.path_formats = [
-        (PF_KEY_DEFAULT, Template("$albumartist/$album%lower{}/$track $title"))
+        (PF_KEY_DEFAULT, "$albumartist/$album%lower{}/$track $title")
     ]
     if prof:
         cProfile.runctx(
diff --git a/beetsplug/deezer.py b/beetsplug/deezer.py
index 65d98f5d0..6ea172631 100644
--- a/beetsplug/deezer.py
+++ b/beetsplug/deezer.py
@@ -215,9 +215,20 @@ class DeezerPlugin(SearchApiMetadataSourcePlugin[IDResponse]):
         name: str,
         va_likely: bool,
     ) -> tuple[str, dict[str, str]]:
-        query = f'album:"{name}"' if query_type == "album" else name
-        if query_type == "track" or not va_likely:
-            query += f' artist:"{artist}"'
+        if query_type == "album":
+            query = f'album:"{name}"'
+            if not va_likely:
+                query += f' artist:"{artist}"'
+        else:
+            # Deezer drops unquoted free text as soon as the query carries any
+            # field:"value" filter, so `<title> artist:"<artist>"` degenerated
+            # into "every track by this artist", truncated to `search_limit`.
+            # The wanted track routinely fell outside that window. Filtering on
+            # the title instead is no better, because `artist:` is fuzzy enough
+            # to match unrelated artists ("Pan Da Punk" for "Daft Punk"), so the
+            # two filters can intersect to nothing even for a well-tagged file.
+            # Plain free text lets Deezer's own relevance ranking do the work.
+            query = f"{name} {artist}".strip()
 
         return query, {}
 
diff --git a/beetsplug/smartplaylist.py b/beetsplug/smartplaylist.py
index 9b0ad6199..77440ac5f 100644
--- a/beetsplug/smartplaylist.py
+++ b/beetsplug/smartplaylist.py
@@ -389,7 +389,7 @@ class SmartPlaylistPlugin(plugins.BeetsPlugin):
             # the items and generate the correct m3u file names.
             matched_items: list[Item] = []
             for item in items:
-                m3u_name = item.evaluate_template(name, True)
+                m3u_name = item.evaluate_template(name, for_path=True)
                 m3u_name = sanitize_path(m3u_name, lib.replacements)
                 item_uri = self.get_item_uri(item)
 
diff --git a/docs/changelog.rst b/docs/changelog.rst
index 1bea07a6d..0ef243968 100644
--- a/docs/changelog.rst
+++ b/docs/changelog.rst
@@ -45,6 +45,13 @@ Bug fixes
   Field names are lowercased when a query is parsed, so such attributes could
   never be matched; flexible attribute lookups now fall back to a
   case-insensitive key match. :bug:`4565`
+- :doc:`plugins/deezer`: Singleton searches now use plain free text rather than
+  ``<title> artist:"<artist>"``. Deezer discards unquoted free text as soon as a
+  query contains any ``field:"value"`` filter, so the old query was evaluated as
+  ``artist:"<artist>"`` alone -- every track by the artist, in Deezer's own
+  relevance order and truncated to ``search_limit``. For artists with more
+  releases than that window, the track being imported was never among the
+  candidates offered.
 
 ..
     For plugin developers
diff --git a/test/plugins/test_deezer.py b/test/plugins/test_deezer.py
new file mode 100644
index 000000000..e62514824
--- /dev/null
+++ b/test/plugins/test_deezer.py
@@ -0,0 +1,41 @@
+import pytest
+
+from beets.library import Item
+from beetsplug.deezer import DeezerPlugin
+
+
+@pytest.fixture
+def plugin():
+    return DeezerPlugin()
+
+
+class TestSearchQuery:
+    def test_track_query_is_free_text(self, plugin):
+        query, filters = plugin.get_search_query_with_filters(
+            "track", [Item()], "Artist", "Title", False
+        )
+
+        assert query == "Title Artist"
+        assert filters == {}
+
+    def test_track_query_tolerates_missing_artist(self, plugin):
+        query, _ = plugin.get_search_query_with_filters(
+            "track", [Item()], "", "Title", False
+        )
+
+        assert query == "Title"
+
+    def test_album_query_filters_on_album_and_artist(self, plugin):
+        query, filters = plugin.get_search_query_with_filters(
+            "album", [Item()], "Artist", "Album", False
+        )
+
+        assert query == 'album:"Album" artist:"Artist"'
+        assert filters == {}
+
+    def test_album_query_omits_artist_for_various_artists(self, plugin):
+        query, _ = plugin.get_search_query_with_filters(
+            "album", [Item()], "Various Artists", "Album", True
+        )
+
+        assert query == 'album:"Album"'
diff --git a/test/plugins/test_smartplaylist.py b/test/plugins/test_smartplaylist.py
index 1d03d6b08..40512049d 100644
--- a/test/plugins/test_smartplaylist.py
+++ b/test/plugins/test_smartplaylist.py
@@ -168,7 +168,7 @@ class SmartPlaylistTest(PlaylistDirMixin, BeetsTestCase):
         spl = SmartPlaylistPlugin()
 
         i = Mock(path=b"/tagada.mp3")
-        i.evaluate_template.side_effect = lambda pl, *_: os.fsdecode(
+        i.evaluate_template.side_effect = lambda pl, **__: os.fsdecode(
             pl
         ).replace("$title", "ta:ga:da")
 
@@ -203,7 +203,7 @@ class SmartPlaylistTest(PlaylistDirMixin, BeetsTestCase):
         type(i).title = PropertyMock(return_value="fake title")
         type(i).length = PropertyMock(return_value=300.123)
         type(i).path = PropertyMock(return_value=b"/tagada.mp3")
-        i.evaluate_template.side_effect = lambda pl, *_: os.fsdecode(
+        i.evaluate_template.side_effect = lambda pl, **__: os.fsdecode(
             pl
         ).replace("$title", "ta:ga:da")
 
@@ -246,7 +246,7 @@ class SmartPlaylistTest(PlaylistDirMixin, BeetsTestCase):
         type(i).path = PropertyMock(return_value=b"/tagada.mp3")
         a = {"id": 456, "genres": ["Rock", "Pop"]}
         i.__getitem__.side_effect = a.__getitem__
-        i.evaluate_template.side_effect = lambda pl, *_: os.fsdecode(
+        i.evaluate_template.side_effect = lambda pl, **__: os.fsdecode(
             pl
         ).replace("$title", "ta:ga:da")
 
diff --git a/test/ui/test_ui.py b/test/ui/test_ui.py
index 065559b0c..3b45b1a1e 100644
--- a/test/ui/test_ui.py
+++ b/test/ui/test_ui.py
@@ -124,9 +124,7 @@ class ConfigTest(IOMixin, TestPluginTestCase):
             config.write("paths: {x: y}")
 
         self.run_command("test")
-        key, template = self.test_cmd.lib.path_formats[0]
-        assert key == "x"
-        assert template.original == "y"
+        assert self.test_cmd.lib.path_formats[0] == ("x", "y")
 
     def test_nonexistant_db(self):
         with self.write_config_file() as config:
diff --git a/test/util/test_pathformats.py b/test/util/test_pathformats.py
index 47846c191..769377a87 100644
--- a/test/util/test_pathformats.py
+++ b/test/util/test_pathformats.py
@@ -6,9 +6,7 @@ def test_get_path_formats(config):
     # override the default 'singleton' path and add a new one
     config["paths"].set({"singleton": "bar", "new": "hello"})
 
-    path_formats = get_path_formats(config["paths"])
-    actual_path_formats = [(key, tmpl.original) for key, tmpl in path_formats]
-    assert actual_path_formats == [
+    assert get_path_formats(config["paths"]) == [
         ("singleton:true", "bar"),
         ("new", "hello"),
         # defaults