langchain-ai/deepagents · error · ModelConfigError

Could not apply {label} to model '{model_name}': {exc}. The

Error message

Could not apply {label} to model '{model_name}': {exc}. The model may not support profile assignment.

What it means

Raised (when `raise_on_failure=True`) by `_apply_profile_overrides` if assigning the merged profile dict onto `model.profile` fails with AttributeError/TypeError/ValueError, meaning the chat model object does not accept profile assignment.

Source

Thrown at libs/code/deepagents_code/config.py:5631

    Raises:
        ModelConfigError: If `raise_on_failure` is `True` and the model
            rejects profile assignment.
    """
    from deepagents_code.model_config import ModelConfigError

    logger.debug("Applying %s profile overrides: %s", label, overrides)
    profile = getattr(model, "profile", None)
    merged = {**profile, **overrides} if isinstance(profile, dict) else overrides
    try:
        model.profile = merged  # ty: ignore[invalid-assignment]
    except (AttributeError, TypeError, ValueError) as exc:
        if raise_on_failure:
            msg = (
                f"Could not apply {label} to model '{model_name}': {exc}. "
                f"The model may not support profile assignment."
            )
            raise ModelConfigError(msg) from exc
        logger.warning(
            "Could not apply %s profile overrides to model '%s': %s. "
            "Overrides will be ignored.",
            label,
            model_name,
            exc,
        )


def create_model(
    model_spec: str | None = None,
    *,
    extra_kwargs: dict[str, Any] | None = None,
    profile_overrides: dict[str, Any] | None = None,
    cli_max_retries: int | None = None,
) -> ModelResult:
    """Create a chat model.

View on GitHub (pinned to a1af029e6e)

Solutions

Set `raise_on_failure=False` (the default) so overrides are ignored with a warning instead of failing model creation.
Use a model class that supports mutable `profile` assignment (standard langchain chat models do).
Drop the profile override for that model, or apply the capability flags another way.
Check the chained exception for which attribute/assignment failed.

Example fix

// before
_apply_profile_overrides(model, overrides, name, label="cli", raise_on_failure=True)
// after
_apply_profile_overrides(model, overrides, name, label="cli", raise_on_failure=False)  # warn + ignore instead

Defensive patterns

Strategy: fallback

Validate before calling

profile = getattr(model, "profile", None)
can_assign = isinstance(profile, dict) or (profile is None and not hasattr(type(model), "__slots__"))
# only request raise_on_failure=True when can_assign

Try / catch

try:
    _apply_profile_overrides(model, overrides, name, label=label, raise_on_failure=True)
except ModelConfigError:
    logger.warning("Profile overrides unsupported for %s; continuing without them", name)

Prevention

Leave raise_on_failure=False (default) unless override application is mandatory
Target profile overrides only at standard langchain chat models
Verify the custom model class has a writable profile attribute before enabling overrides
Confirm the langchain version's profile schema matches your override keys

When it happens

Trigger: `create_model` with profile overrides (config.toml or `--profile-override`) applied to a custom `BaseChatModel` whose `profile` attribute is read-only, absent in a way that rejects setattr, or validated against a strict schema.

Common situations: Third-party or custom model classes with frozen/slot-based attributes, provider integrations where `profile` is a property without a setter, langchain version where profile shape changed.

Related errors

AI-assisted analysis of langchain-ai/deepagents@a1af029e6e (2026-08-29). Data as JSON: /api/errors/014da6e13e0790df. Report an issue: GitHub.