diff --git a/litellm/litellm_core_utils/token_counter.py b/litellm/litellm_core_utils/token_counter.py
index c766c6edec1..caa3b9d4faa 100644
--- a/litellm/litellm_core_utils/token_counter.py
+++ b/litellm/litellm_core_utils/token_counter.py
@@ -724,27 +724,29 @@ def _count_content_list(
         for c in content_list:
             if isinstance(c, str):
                 num_tokens += count_function(c)
-            elif c["type"] == "text":
+                continue
+            content_type = c.get("type")
+            if content_type == "text":
                 num_tokens += count_function(str(c.get("text", "")))
-            elif c["type"] == "image_url":
+            elif content_type == "image_url":
                 image_url = c.get("image_url")
                 num_tokens += _count_image_tokens(
                     image_url, use_default_image_token_count
                 )
-            elif c["type"] in ("tool_use", "tool_result"):
+            elif content_type in ("tool_use", "tool_result"):
                 num_tokens += _count_anthropic_content(
                     c,
                     count_function,
                     use_default_image_token_count,
                     default_token_count,
                 )
-            elif c["type"] == "thinking":
+            elif content_type == "thinking":
                 # Claude extended thinking content block
                 # Count the thinking text and skip signature (opaque signature blob)
                 thinking_text = str(c.get("thinking", ""))
                 if thinking_text:
                     num_tokens += count_function(thinking_text)
-            elif c["type"] == "tool_reference":
+            elif content_type == "tool_reference":
                 # Anthropic tool-search reference block: a lightweight pointer to
                 # a deferred tool, e.g. {"type": "tool_reference", "tool_name": ...}.
                 # The full tool definition is counted via the `tools` param, so we
@@ -755,16 +757,6 @@ def _count_content_list(
                 tool_name = str(c.get("tool_name") or "")
                 if tool_name:
                     num_tokens += count_function(tool_name)
-            else:
-                content_type = (
-                    c.get("type", type(c).__name__)
-                    if isinstance(c, dict)
-                    else type(c).__name__
-                )
-                raise ValueError(
-                    f"Invalid content item type: {content_type}. "
-                    f"Expected str or dict with 'type' field (text, image_url, tool_use, tool_result, thinking, tool_reference)."
-                )
         return num_tokens
     except Exception as e:
         if default_token_count is not None:
diff --git a/tests/test_litellm/litellm_core_utils/test_token_counter.py b/tests/test_litellm/litellm_core_utils/test_token_counter.py
index 60e5a797627..cca4ed14798 100644
--- a/tests/test_litellm/litellm_core_utils/test_token_counter.py
+++ b/tests/test_litellm/litellm_core_utils/test_token_counter.py
@@ -1051,22 +1051,52 @@ def test_token_counter_with_tool_reference_block():
     assert tokens_empty >= 0
 
 
-def test_count_content_list_rejects_unknown_type():
+def test_count_content_list_skips_unknown_type():
     """
-    An unrecognized content block type must raise, and the error message must
-    enumerate the supported types (including `tool_reference`). This pins the
-    catch-all contract so a future block type isn't silently dropped.
+    Regression test: a non-standard content block (e.g. a custom block emitted
+    by a tool orchestrator) must NOT crash the token counter.
+
+    Before the fix, _count_content_list raised
+    `Invalid content item type: ...` on any unrecognized block, which bubbled
+    up as `Error getting number of tokens from content list` and broke token
+    counting for the whole message. The counter must instead skip the unknown
+    block while still counting recognized siblings.
     """
     from litellm.litellm_core_utils.token_counter import _count_content_list
 
-    with pytest.raises(ValueError) as exc_info:
-        _count_content_list(
-            count_function=len,
-            content_list=[{"type": "totally_unknown_block"}],
-            use_default_image_token_count=False,
-            default_token_count=None,
-        )
+    long_text = "the quick brown fox jumps over the lazy dog " * 20
+    text_only_tokens = _count_content_list(
+        count_function=len,
+        content_list=[{"type": "text", "text": long_text}],
+        use_default_image_token_count=False,
+        default_token_count=None,
+    )
+    assert text_only_tokens > 0
 
-    message = str(exc_info.value)
-    assert "Invalid content item type: totally_unknown_block" in message
-    assert "tool_reference" in message
+    with_unknown_tokens = _count_content_list(
+        count_function=len,
+        content_list=[
+            {"type": "text", "text": long_text},
+            {"type": "totally_unknown_block", "payload": {"k": "v"}},
+        ],
+        use_default_image_token_count=False,
+        default_token_count=None,
+    )
+    assert with_unknown_tokens == text_only_tokens
+
+    only_unknown_tokens = _count_content_list(
+        count_function=len,
+        content_list=[{"type": "weird_custom_block", "blob": "x"}],
+        use_default_image_token_count=False,
+        default_token_count=None,
+    )
+    assert only_unknown_tokens == 0
+
+    # A dict block with no `type` key at all must also be skipped, not crash.
+    no_type_tokens = _count_content_list(
+        count_function=len,
+        content_list=[{"text": "ignored because there is no type field"}],
+        use_default_image_token_count=False,
+        default_token_count=None,
+    )
+    assert no_type_tokens == 0