Anthropic automatic caching with ChatAnthropicVertex throws exception

[henriquejsb]

Package (Required)

• langchain-google-genai
• langchain-google-vertexai
• langchain-google-community
• Other / not sure / general

Checked other resources

• I added a descriptive title to this issue
• I searched the LangChain documentation and API reference (linked above)
• I used the GitHub search to find a similar issue and didn't find it
• I am sure this is a bug and not a question or request for help

Example Code (Python)

import os

from dotenv import load_dotenv
from langchain_anthropic import ChatAnthropic
from langchain_core.messages import AIMessage, HumanMessage, SystemMessage
from langchain_google_vertexai.model_garden import ChatAnthropicVertex

load_dotenv()

project = os.getenv("GOOGLE_CLOUD_PROJECT")
location = os.getenv("GOOGLE_CLOUD_LOCATION")

# Name: langchain-google-vertexai
# Version: 3.2.2
# Name: anthropic
# Version: 0.83.0
# Name: langchain
# Version: 1.2.10
# Name: langchain-anthropic
# Version: 1.3.3
model = ChatAnthropicVertex(
    name="claude-sonnet-4-6",
    model_name="claude-sonnet-4-6",
    project=project,
    location=location,
    temperature=0.7,
    max_output_tokens=8192,
    max_retries=1,
    # additional_headers={"anthropic-beta": "extended-cache-ttl-2025-04-11"},
)

system_message = SystemMessage(content="You are a helpful assistant.")
human_message = HumanMessage(content="What is the capital of France?")
messages = [system_message, human_message]
print("Trying with ChatAnthropicVertex...")
try:
    response = model.invoke(messages, cache_control={"type": "ephemeral"})
    print(response.text)
except Exception as e:
    print(f"Error during Anthropic Vertex model invocation: {e}")

# With Anthropic
model = ChatAnthropic(
    name="claude-sonnet-4-6",
    model_name="claude-sonnet-4-6",
    temperature=0.7,
    max_tokens=8192,
)
print("Trying with ChatAnthropic...")
try:
    response = model.invoke(messages, cache_control={"type": "ephemeral"})
    print(response.text)
except Exception as e:
    print(f"Error during Anthropic model invocation: {e}")


"""
OUTPUT:
Trying with ChatAnthropicVertex...
Error during Anthropic Vertex model invocation: Error code: 400 - {'type': 'error', 'error': {'type': 'invalid_request_error', 'message': 'cache_control: Extra inputs are not permitted'}, 'request_id': 'req_vrtx_011CYSk27BXt381h7s6vN6HV'}
Trying with ChatAnthropic...
The capital of France is **Paris**.
"""

Error Message and Stack Trace (if applicable)

Traceback (most recent call last):
  File "/Users/someuser/example-proj/test_bug_vertex.py", line 40, in <module>
    raise e
  File "/Users/someuser/example-proj/test_bug_vertex.py", line 37, in <module>
    response = model.invoke(messages, cache_control={"type": "ephemeral"})
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/langchain_core/language_models/chat_models.py", line 402, in invoke
    self.generate_prompt(
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/langchain_core/language_models/chat_models.py", line 1123, in generate_prompt
    return self.generate(prompt_messages, stop=stop, callbacks=callbacks, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/langchain_core/language_models/chat_models.py", line 933, in generate
    self._generate_with_cache(
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/langchain_core/language_models/chat_models.py", line 1235, in _generate_with_cache
    result = self._generate(
             ^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/langchain_google_vertexai/model_garden.py", line 343, in _generate
    data = _completion_with_retry_inner(**params)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/tenacity/__init__.py", line 331, in wrapped_f
    return copy(f, *args, **kw)
           ^^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/tenacity/__init__.py", line 470, in __call__
    do = self.iter(retry_state=retry_state)
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/tenacity/__init__.py", line 371, in iter
    result = action(retry_state)
             ^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/tenacity/__init__.py", line 413, in exc_check
    raise retry_exc.reraise()
          ^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/tenacity/__init__.py", line 184, in reraise
    raise self.last_attempt.result()
          ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/concurrent/futures/_base.py", line 449, in result
    return self.__get_result()
           ^^^^^^^^^^^^^^^^^^^
  File "/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/concurrent/futures/_base.py", line 401, in __get_result
    raise self._exception
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/tenacity/__init__.py", line 473, in __call__
    result = fn(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/langchain_google_vertexai/model_garden.py", line 341, in _completion_with_retry_inner
    return self.client.messages.create(**params)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/anthropic/_utils/_utils.py", line 282, in wrapper
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/anthropic/resources/messages/messages.py", line 996, in create
    return self._post(
           ^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/anthropic/_base_client.py", line 1364, in post
    return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/someuser/example-proj/.venv/lib/python3.11/site-packages/anthropic/_base_client.py", line 1137, in request
    raise self._make_status_error_from_response(err.response) from None
anthropic.BadRequestError: Error code: 400 - {'type': 'error', 'error': {'type': 'invalid_request_error', 'message': 'cache_control: Extra inputs are not permitted'}, 'request_id': 'req_vrtx_011CYSk5seYzVnbvX4vhA5Ky'}

Description

I'm trying to use Anthropic's automatic prompt caching.
According to Langchain, I can simply add a "cache_control" extra kwarg when invoking the model.
This works when using ChatAnthropic from langchain-anthropic, but raises an error when using ChatAnthropicVertex.