diff --git a/litellm/llms/mistral/chat/transformation.py b/litellm/llms/mistral/chat/transformation.py
index 0c95fd4df07..8259706d84c 100644
--- a/litellm/llms/mistral/chat/transformation.py
+++ b/litellm/llms/mistral/chat/transformation.py
@@ -6,7 +6,8 @@ Why separate file? Make it easy to see how transformation works
 Docs - https://docs.mistral.ai/api/
 """
 
-from collections.abc import AsyncIterator, Coroutine, Iterator
+from collections.abc import AsyncIterator, Coroutine, Iterator, Mapping
+from types import MappingProxyType
 from typing import Any, Final, Literal, cast, get_type_hints, overload
 
 import httpx
@@ -24,7 +25,7 @@ from litellm.secret_managers.main import get_secret_str
 from litellm.types.llms.mistral import MistralThinkingBlock, MistralToolCallMessage
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.utils import ModelResponse, ModelResponseStream
-from litellm.utils import convert_to_model_response_object
+from litellm.utils import convert_to_model_response_object, supports_reasoning
 
 
 class MistralConfig(OpenAIGPTConfig):
@@ -98,12 +99,26 @@ class MistralConfig(OpenAIGPTConfig):
             "parallel_tool_calls",
         ]
 
-        # Add reasoning support for magistral models
-        if "magistral" in model.lower():
+        if self._is_magistral_model(model):
             supported_params.extend(["thinking", "reasoning_effort"])
+        elif self._supports_native_reasoning_effort(model):
+            supported_params.append("reasoning_effort")
 
         return supported_params
 
+    @staticmethod
+    def _is_magistral_model(model: str) -> bool:
+        return "magistral" in model.lower()
+
+    @staticmethod
+    def _supports_native_reasoning_effort(model: str) -> bool:
+        """
+        Mistral's hybrid reasoning models (mistral-medium-3-5 and newer) take ``reasoning_effort``
+        directly on /chat/completions, so it is forwarded as-is instead of being emulated with the
+        magistral system prompt: https://docs.mistral.ai/capabilities/reasoning
+        """
+        return supports_reasoning(model=model, custom_llm_provider="mistral")
+
     def _map_tool_choice(self, tool_choice: str) -> str:
         if tool_choice == "auto" or tool_choice == "none":
             return tool_choice
@@ -168,16 +183,19 @@ class MistralConfig(OpenAIGPTConfig):
                 optional_params["extra_body"] = {"random_seed": value}
             if param == "response_format":
                 optional_params["response_format"] = value
-            if param == "reasoning_effort" and "magistral" in model.lower():
-                # Flag that we need to add reasoning system prompt
-                optional_params["_add_reasoning_prompt"] = True
-            if param == "thinking" and "magistral" in model.lower():
-                # Flag that we need to add reasoning system prompt
-                optional_params["_add_reasoning_prompt"] = True
+            if param in ("reasoning_effort", "thinking"):
+                optional_params.update(self._map_reasoning_param(param=param, value=value, model=model))
             if param == "parallel_tool_calls":
                 optional_params["parallel_tool_calls"] = value
         return optional_params
 
+    def _map_reasoning_param(self, param: str, value: object, model: str) -> Mapping[str, object]:
+        if self._is_magistral_model(model):
+            return MappingProxyType({"_add_reasoning_prompt": True})
+        if param == "reasoning_effort" and value is not None and self._supports_native_reasoning_effort(model):
+            return MappingProxyType({"reasoning_effort": value})
+        return MappingProxyType({})
+
     def _get_openai_compatible_provider_info(self, api_base: str | None, api_key: str | None) -> tuple[str, str | None]:
         # mistral is openai compatible, we just need to set this to custom_openai and have the api_base be https://api.mistral.ai
         api_base = (
diff --git a/tests/test_litellm/llms/mistral/test_mistral_chat_transformation.py b/tests/test_litellm/llms/mistral/test_mistral_chat_transformation.py
index 15694d9f218..ae72a74d3ae 100644
--- a/tests/test_litellm/llms/mistral/test_mistral_chat_transformation.py
+++ b/tests/test_litellm/llms/mistral/test_mistral_chat_transformation.py
@@ -307,6 +307,48 @@ class TestMistralReasoningSupport:
         assert len(result["messages"]) == 1
         assert result["messages"][0]["role"] == "user"
 
+    @pytest.mark.parametrize(
+        "model",
+        ["mistral/mistral-medium-3-5", "mistral-medium-3-5", "mistral/mistral-medium-latest"],
+    )
+    def test_native_reasoning_model_supports_reasoning_effort(self, model):
+        """Hybrid reasoning models take reasoning_effort natively, see https://github.com/BerriAI/litellm/issues/36407"""
+        supported_params = MistralConfig().get_supported_openai_params(model)
+
+        assert "reasoning_effort" in supported_params
+        assert "thinking" not in supported_params
+
+    def test_native_reasoning_effort_forwarded_to_request_body(self):
+        mistral_config = MistralConfig()
+
+        optional_params = mistral_config.map_openai_params(
+            non_default_params={"reasoning_effort": "high"},
+            optional_params={},
+            model="mistral/mistral-medium-3-5",
+            drop_params=False,
+        )
+        assert optional_params == {"reasoning_effort": "high"}
+
+        request = mistral_config.transform_request(
+            model="mistral/mistral-medium-3-5",
+            messages=[{"role": "user", "content": "What is 15 * 7?"}],
+            optional_params=optional_params,
+            litellm_params={},
+            headers={},
+        )
+        assert request["reasoning_effort"] == "high"
+        assert len(request["messages"]) == 1
+
+    def test_magistral_reasoning_effort_stays_prompt_based(self):
+        optional_params = MistralConfig().map_openai_params(
+            non_default_params={"reasoning_effort": "high"},
+            optional_params={},
+            model="mistral/magistral-medium-2506",
+            drop_params=False,
+        )
+
+        assert optional_params == {"_add_reasoning_prompt": True}
+
     def test_case_insensitive_magistral_detection(self):
         """Test that magistral model detection is case-insensitive."""
         mistral_config = MistralConfig()