invoke-ai/InvokeAI · error · DuplicateModelException

More than one model matched the search criteria: base_model=

Error message

More than one model matched the search criteria: base_model='{base_model}', model_type='{model_type}', model_name='{model_name}'.

What it means

model_info_by_name looks up a model by name/base/type via search_by_attr and requires exactly one match. When more than one installed model config matches all three criteria it raises DuplicateModelException, because it cannot disambiguate which record to return.

Solutions

1. Call search_by_attr yourself, list the duplicate keys, and delete one via del_model.
2. Rename one of the duplicate models so names are unique.
3. Uninstall the stale/duplicate copy through the Model Manager UI.
4. Reference the model by its unique key instead of name where the API allows.

Example fix

# before
info = records.model_info_by_name('MyModel', BaseModelType.StableDiffusion1, ModelType.Main)
# after
configs = records.search_by_attr(model_name='MyModel', base_model=..., model_type=...)
for c in configs[1:]:
    records.del_model(c.key)
info = records.model_info_by_name('MyModel', BaseModelType.StableDiffusion1, ModelType.Main)

configs = records.search_by_attr(model_name=name, base_model=base, model_type=mtype)
assert len(configs) <= 1, f"Duplicates present: {[c.key for c in configs]}"

try:
    info = records.model_info_by_name(name, base, mtype)
except DuplicateModelException as e:
    configs = records.search_by_attr(model_name=name, base_model=base, model_type=mtype)
    info = min(configs, key=lambda c: c.key)  # or prompt user to disambiguate
    log.warning('Duplicate models for %s; picking %s', name, info.key)

When it happens

Trigger: Calling model_info_by_name (or API endpoints using it) where two installed models share the same name, base_model and model_type — e.g. the same model installed twice from different paths/sources, or after a reinstall that didn't dedupe.

Common situations: Re-importing the same checkpoint from a new folder without deleting the old entry; scanning multiple model directories containing copies of the same file; merged/fine-tuned variants saved under an identical name in the same base/type.

Related errors

• A model with path ' ' is already installed
• LoRA " " already applied to CLIP encoder.
• LoRA " " already applied to T5 encoder.
• LoRA " " already applied to transformer.
• source_url must be a string

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/6b545f9176f79058. Report an issue: GitHub.

Appendix: source

Thrown at invokeai/app/services/model_records/model_records_base.py:300

        If none of the optional filters are passed, will return all
        models in the database.
        """
        pass

    def all_models(self) -> List[AnyModelConfig]:
        """Return all the model configs in the database."""
        return self.search_by_attr()

    def model_info_by_name(self, model_name: str, base_model: BaseModelType, model_type: ModelType) -> AnyModelConfig:
        """
        Return information about a single model using its name, base type and model type.

        If there are more than one model that match, raises a DuplicateModelException.
        If no model matches, raises an UnknownModelException
        """
        model_configs = self.search_by_attr(model_name=model_name, base_model=base_model, model_type=model_type)
        if len(model_configs) > 1:
            raise DuplicateModelException(
                f"More than one model matched the search criteria: base_model='{base_model}', model_type='{model_type}', model_name='{model_name}'."
            )
        if len(model_configs) == 0:
            raise UnknownModelException(
                f"More than one model matched the search criteria: base_model='{base_model}', model_type='{model_type}', model_name='{model_name}'."
            )
        return model_configs[0]

View on GitHub (pinned to 0b6a024f2f)