mem0ai/mem0 · error · ImportError

The 'anthropic' library is required. Please install it using

Error message

The 'anthropic' library is required. Please install it using 'pip install anthropic'.

What it means

Module-level ImportError raised when mem0/llms/anthropic.py is imported without the `anthropic` package installed. Import happens lazily when mem0 instantiates AnthropicLLM (llm provider 'anthropic'), so the error surfaces at Memory creation or first config load, not at pip install time.

Source

Thrown at mem0/llms/anthropic.py:7

import os
from typing import Dict, List, Optional, Union

try:
    import anthropic
except ImportError:
    raise ImportError("The 'anthropic' library is required. Please install it using 'pip install anthropic'.")

from mem0.configs.llms.anthropic import AnthropicConfig
from mem0.configs.llms.base import BaseLlmConfig
from mem0.llms.base import LLMBase


class AnthropicLLM(LLMBase):
    def __init__(self, config: Optional[Union[BaseLlmConfig, AnthropicConfig, Dict]] = None):
        # Convert to AnthropicConfig if needed
        if config is None:
            config = AnthropicConfig()
        elif isinstance(config, dict):
            config = AnthropicConfig(**config)
        elif isinstance(config, BaseLlmConfig) and not isinstance(config, AnthropicConfig):
            # Convert BaseLlmConfig to AnthropicConfig
            config = AnthropicConfig(
                model=config.model,
                temperature=config.temperature,

View on GitHub (pinned to 001c235229)

Solutions

pip install anthropic (or add mem0ai[anthropic]-style extras / declare anthropic in requirements)
Confirm the install is in the same environment: python -m pip install anthropic using the interpreter that runs mem0
Verify with python -c "import anthropic; print(anthropic.__version__)"
If it still fails, check for a local file/dir named anthropic.py shadowing the package

Example fix

// before
Memory.from_config({"llm": {"provider": "anthropic", "config": {"model": "claude-sonnet-4-5"}}})  # ImportError

# after
# pip install anthropic
Memory.from_config({"llm": {"provider": "anthropic", "config": {"model": "claude-sonnet-4-5", "api_key": "..."}}})

Defensive patterns

Strategy: validation

Validate before calling

import importlib.util

if importlib.util.find_spec("anthropic") is None:
    raise SystemExit("Missing dependency: pip install anthropic")

Try / catch

try:
    from mem0.llms.anthropic import AnthropicLLM
except ImportError as e:
    raise SystemExit(f"{e}; install with: pip install anthropic") from e

Prevention

Declare anthropic in requirements when llm.provider is anthropic
Add a startup smoke test that imports every configured provider
Beware file shadowing: never name a local module anthropic.py

When it happens

Trigger: Setting llm.provider to anthropic in config without pip install anthropic; deploying mem0 in a slim container with only core deps; CI lockfile missing the optional dependency.

Common situations: Trying Claude models after only installing mem0ai base; venv mismatch where anthropic is installed in a different interpreter; dependency conflicts (anthropic pinned/removed by another package).

Related errors

AI-assisted analysis of mem0ai/mem0@001c235229 (2026-08-15). Data as JSON: /api/errors/f2423d1ddb39bd74. Report an issue: GitHub.