agentscope-ai/agentscope · error · TypeError

f"AgentScopeLLM `model` must be a ChatModelBase, got {type(s

Error message

f"AgentScopeLLM `model` must be a ChatModelBase, got {type(self.config.model).__name__}."

What it means

AgentScopeLLM validates that config.model is an instance of AgentScope's ChatModelBase. Passing a string name, a LangChain/OpenAI client, or any other object raises this TypeError naming the offending type.

Source

Thrown at src/agentscope/middleware/_longterm_memory/_mem0/_agentscope_adapter.py:115

    Pass your AgentScope model into ``config["model"]``; mem0's memory
    extraction calls then route through it. Both streaming and
    non-streaming AgentScope models are accepted (streaming responses are
    drained and the final chunk is used).
    """

    def __init__(
        self,
        config: BaseLlmConfig | dict | None = None,
    ) -> None:
        """Initialize the AgentScope LLM for mem0."""
        super().__init__(config)
        if self.config.model is None:
            raise ValueError(
                "AgentScopeLLM requires `model` in the config to be an "
                "AgentScope ChatModelBase instance.",
            )
        if not isinstance(self.config.model, ChatModelBase):
            raise TypeError(
                f"AgentScopeLLM `model` must be a ChatModelBase, got "
                f"{type(self.config.model).__name__}.",
            )
        self._agentscope_model: ChatModelBase = self.config.model
        self._bridge = _AsyncBridge()

    # ----- LLMBase interface -----
    # pylint: disable=unused-argument
    def generate_response(
        self,
        messages: list[dict[str, str]],
        response_format: Any | None = None,  # mem0 contract — unused
        tools: list[dict] | None = None,
        tool_choice: str = "auto",  # mem0 contract — unused
    ) -> str | dict:
        """mem0 ``LLMBase`` entry — runs the AgentScope chat model
        synchronously and returns str (or dict with tool_calls when
        ``tools`` is given)."""

View on GitHub (pinned to e90f1c7592)

Solutions

Wrap the provider in an AgentScope model class (OpenAIChat, DashScopeChat, etc.) and pass that instance
Check the object came from agentscope.model, not the raw provider SDK
Let build_mem0_config / Mem0Middleware construct the adapter from chat_model=

Example fix

// before
AgentScopeLLM({'model': 'gpt-4o', 'provider': 'agentscope'})
// after
from agentscope.model import OpenAIChat
AgentScopeLLM({'model': OpenAIChat(model='gpt-4o')})

Defensive patterns

Strategy: type-guard

Validate before calling

from agentscope.model import ChatModelBase
if not isinstance(config.get('model'), ChatModelBase):
    raise TypeError('model must be ChatModelBase, e.g. OpenAIChat(...)')

Type guard

from agentscope.model import ChatModelBase
def is_chat_model_base(m) -> bool:
    return isinstance(m, ChatModelBase)

Try / catch

try:
    llm = AgentScopeLLM(cfg)
except TypeError as e:
    raise SystemExit(f'Bad model config: {e}') from e

Prevention

Never pass model names as strings to agentscope adapters
Import model classes from agentscope.model, not the provider SDK

When it happens

Trigger: AgentScopeLLM({'model': 'gpt-4o'}) (string), or config.model set to an OpenAI SDK client, a mem0 OpenAILLM, or any non-AgentScope chat wrapper.

Common situations: Assuming mem0's string model naming convention applies; passing an SDK client object instead of an AgentScope model wrapper; mixing adapters between frameworks.

Related errors

AI-assisted analysis of agentscope-ai/agentscope@e90f1c7592 (2026-08-28). Data as JSON: /api/errors/c2d0ef1c8f9104cd. Report an issue: GitHub.