diff --git a/ci_cd/generate_model_prices_schema.py b/ci_cd/generate_model_prices_schema.py
index 359df8e6e32..a5a4b7f6f3b 100644
--- a/ci_cd/generate_model_prices_schema.py
+++ b/ci_cd/generate_model_prices_schema.py
@@ -259,6 +259,11 @@ def cost_schema(key: str) -> JsonSchema:
 
 def string_key_schemas(modes: tuple) -> dict[str, JsonSchema]:
     return {
+        "token_count_tokenizer": {
+            "type": "string",
+            "description": "Tokenizer litellm counts this model's requests with when the provider cannot count them.",
+            "enum": ["kimi_k3"],
+        },
         "litellm_provider": {
             "type": "string",
             "description": "LiteLLM provider slug; one of https://docs.litellm.ai/docs/providers.",
diff --git a/litellm/proxy/proxy_server.py b/litellm/proxy/proxy_server.py
index 4ec1cf36408..8369e79596c 100644
--- a/litellm/proxy/proxy_server.py
+++ b/litellm/proxy/proxy_server.py
@@ -14153,8 +14153,10 @@ async def count_request_tokens(
             model_info.get("custom_tokenizer", None),
         )
     system_message: Final = _system_message(system)
+    prompt_turn: Final = prompt_only and not (system_message is None and not tools)
     typed_messages: Final = cast(  # cast-ok: request messages are raw chat-shaped dicts that token_counter normalizes
-        Sequence[AllMessageValues] | None, messages
+        Sequence[AllMessageValues] | None,
+        [{"role": "user", "content": prompt}] if prompt_turn else (None if prompt_only else messages),
     )
     counted_messages: Final = (
         typed_messages if typed_messages is None or system_message is None else (system_message, *typed_messages)