diff --git a/plugins/platforms/matrix/adapter.py b/plugins/platforms/matrix/adapter.py
index f86a45366f7..a49a99bdd0e 100644
--- a/plugins/platforms/matrix/adapter.py
+++ b/plugins/platforms/matrix/adapter.py
@@ -837,6 +837,12 @@ class MatrixAdapter(BasePlatformAdapter):
     # (see _normalize_matrix_bang_command), so instruction text shows "!".
     typed_command_prefix = "!"
 
+    # Class-level defaults so partially-constructed instances (tests build
+    # adapters via object.__new__ without __init__) keep working; __init__
+    # overrides both from _resolve_max_message_length().
+    max_message_length = DEFAULT_MAX_MESSAGE_LENGTH
+    _split_threshold = DEFAULT_MAX_MESSAGE_LENGTH - 100
+
     def __init__(self, config: PlatformConfig):
         super().__init__(config, Platform.MATRIX)
 
diff --git a/tests/gateway/test_text_batching.py b/tests/gateway/test_text_batching.py
index d72cb439d47..6c0cdaa6d9f 100644
--- a/tests/gateway/test_text_batching.py
+++ b/tests/gateway/test_text_batching.py
@@ -278,9 +278,9 @@ class TestMatrixTextBatching:
 
     @pytest.mark.asyncio
     async def test_adaptive_delay_for_near_limit_chunk(self):
-        """Chunks near the 4000-char limit should trigger longer delay."""
+        """Chunks near the outbound limit should trigger longer delay."""
         adapter = _make_matrix_adapter()
-        long_text = "x" * 3950
+        long_text = "x" * (adapter._split_threshold + 50)
         adapter._enqueue_text_event(_make_event(long_text, Platform.MATRIX))
 
         await asyncio.sleep(0.15)