diff --git a/ollama/_client.py b/ollama/_client.py
index 8dfce82..779b988 100644
--- a/ollama/_client.py
+++ b/ollama/_client.py
@@ -209,7 +209,7 @@ def generate(
     template: str = '',
     context: Optional[Sequence[int]] = None,
     stream: Literal[False] = False,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     raw: bool = False,
@@ -233,7 +233,7 @@ def generate(
     template: str = '',
     context: Optional[Sequence[int]] = None,
     stream: Literal[True] = True,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     raw: bool = False,
@@ -256,7 +256,7 @@ def generate(
     template: Optional[str] = None,
     context: Optional[Sequence[int]] = None,
     stream: bool = False,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     raw: Optional[bool] = None,
@@ -313,7 +313,7 @@ def chat(
     *,
     tools: Optional[Sequence[Union[Mapping[str, Any], Tool, Callable]]] = None,
     stream: Literal[False] = False,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     format: Optional[Union[Literal['', 'json'], JsonSchemaValue]] = None,
@@ -329,7 +329,7 @@ def chat(
     *,
     tools: Optional[Sequence[Union[Mapping[str, Any], Tool, Callable]]] = None,
     stream: Literal[True] = True,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     format: Optional[Union[Literal['', 'json'], JsonSchemaValue]] = None,
@@ -344,7 +344,7 @@ def chat(
     *,
     tools: Optional[Sequence[Union[Mapping[str, Any], Tool, Callable]]] = None,
     stream: bool = False,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     format: Optional[Union[Literal['', 'json'], JsonSchemaValue]] = None,
@@ -842,7 +842,7 @@ async def generate(
     template: str = '',
     context: Optional[Sequence[int]] = None,
     stream: Literal[False] = False,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     raw: bool = False,
@@ -866,7 +866,7 @@ async def generate(
     template: str = '',
     context: Optional[Sequence[int]] = None,
     stream: Literal[True] = True,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     raw: bool = False,
@@ -889,7 +889,7 @@ async def generate(
     template: Optional[str] = None,
     context: Optional[Sequence[int]] = None,
     stream: bool = False,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     raw: Optional[bool] = None,
@@ -945,7 +945,7 @@ async def chat(
     *,
     tools: Optional[Sequence[Union[Mapping[str, Any], Tool, Callable]]] = None,
     stream: Literal[False] = False,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     format: Optional[Union[Literal['', 'json'], JsonSchemaValue]] = None,
@@ -961,7 +961,7 @@ async def chat(
     *,
     tools: Optional[Sequence[Union[Mapping[str, Any], Tool, Callable]]] = None,
     stream: Literal[True] = True,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     format: Optional[Union[Literal['', 'json'], JsonSchemaValue]] = None,
@@ -976,7 +976,7 @@ async def chat(
     *,
     tools: Optional[Sequence[Union[Mapping[str, Any], Tool, Callable]]] = None,
     stream: bool = False,
-    think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None,
+    think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None,
     logprobs: Optional[bool] = None,
     top_logprobs: Optional[int] = None,
     format: Optional[Union[Literal['', 'json'], JsonSchemaValue]] = None,
diff --git a/ollama/_types.py b/ollama/_types.py
index 96529d6..b66cb67 100644
--- a/ollama/_types.py
+++ b/ollama/_types.py
@@ -207,7 +207,7 @@ class GenerateRequest(BaseGenerateRequest):
   images: Optional[Sequence[Image]] = None
   'Image data for multimodal models.'
 
-  think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None
+  think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None
   'Enable thinking mode (for thinking models).'
 
   logprobs: Optional[bool] = None
@@ -400,7 +400,7 @@ def serialize_model(self, nxt):
   tools: Optional[Sequence[Tool]] = None
   'Tools to use for the chat.'
 
-  think: Optional[Union[bool, Literal['low', 'medium', 'high']]] = None
+  think: Optional[Union[bool, Literal['low', 'medium', 'high', 'max']]] = None
   'Enable thinking mode (for thinking models).'
 
   logprobs: Optional[bool] = None
diff --git a/tests/test_client.py b/tests/test_client.py
index 7b7ab38..498f152 100644
--- a/tests/test_client.py
+++ b/tests/test_client.py
@@ -1490,10 +1490,28 @@ async def test_async_client_context_manager():
 
 
 def test_generate_think_annotation_matches_chat():
-  # The `think` parameter accepts bool or the 'low'/'medium'/'high' string levels.
+  # The `think` parameter accepts bool or the 'low'/'medium'/'high'/'max' string levels.
   # Client.generate must keep the same annotation as Client.chat and
   # AsyncClient.generate so passing a string level does not raise a false type
   # error (regression guard for the sync generate overloads/implementation).
   expected = inspect.signature(Client.chat).parameters['think'].annotation
   assert inspect.signature(Client.generate).parameters['think'].annotation == expected
   assert inspect.signature(AsyncClient.generate).parameters['think'].annotation == expected
+
+
+def test_think_max_level_accepted():
+  # The 'max' think level must be accepted by all generate/chat methods without
+  # raising a type error (mirrors the Go ThinkValue which accepts 'max').
+  import typing
+
+  for method in (Client.generate, Client.chat, AsyncClient.generate, AsyncClient.chat):
+    annotation = typing.get_type_hints(method).get('think')
+    args = typing.get_args(annotation)
+    # Flatten nested Unions/Literals to find the literal values
+    literals = set()
+    for arg in args:
+      if arg is bool or arg is bool:
+        continue
+      sub = typing.get_args(arg)
+      literals.update(sub)
+    assert 'max' in literals, f'{method.__qualname__} think annotation missing max: {literals}'