invoke-ai/InvokeAI · error · ValidationError

No model config class found for type={target_type!r}

Error message

No model config class found for type={target_type!r}

What it means

_construct_config_for_type tries candidate pydantic config classes for a given model record type; if none exists for target_type and no candidate produced a ValidationError, it synthesizes a ValidationError stating no config class was found. This surfaces through update_model when a model record has an unknown/unregistered model type.

Source

Thrown at invokeai/app/services/model_records/model_records_sql.py:86

def _construct_config_for_type(fields: dict, target_type: ModelType) -> AnyModelConfig:
    """Try every config class whose `type` default matches `target_type` and return the first that validates.

    Used when changing a model's type via the update endpoint: the existing record's `format`/`variant`
    fields belong to the old class and may not have a discriminator match in the new type space, so we
    fall back to constructing each candidate class directly with whatever fields it accepts.
    """
    last_error: Exception | None = None
    for candidate_class in Config_Base.CONFIG_CLASSES:
        type_field = candidate_class.model_fields.get("type")
        if type_field is None or type_field.default != target_type:
            continue
        try:
            return candidate_class(**fields)  # type: ignore[return-value]
        except ValidationError as e:
            last_error = e
    if last_error is not None:
        raise last_error
    raise ValidationError.from_exception_data(
        f"No model config class found for type={target_type!r}",
        line_errors=[],
    )


class ModelRecordServiceSQL(ModelRecordServiceBase):
    """Implementation of the ModelConfigStore ABC using a SQL database."""

    def __init__(self, db: SqliteDatabase, logger: logging.Logger):
        """
        Initialize a new object from preexisting sqlite3 connection and threading lock objects.

        :param db: Sqlite connection object
        """
        super().__init__()
        self._db = db
        self._logger = logger

View on GitHub (pinned to 0b6a024f2f)

Solutions

  1. Check the type field of the offending model row and correct it to a valid model type.
  2. Update InvokeAI so the running code recognizes the record's type (version mismatch).
  3. Delete the invalid record via del_model and re-add the model.
  4. Restore the models DB from a backup or re-scan models from disk.

Example fix

# before
# models table row: {"type": "main_v3_experimental", ...}
# after
# correct the row to a known type
# {"type": "main", ...}  (or upgrade InvokeAI to a version that knows the type)
Defensive patterns

Strategy: try-catch

Validate before calling

KNOWN_TYPES = {'main','vae','controlnet','embedding','lora','t2i_adapter','textual_inversion','ip_adapter','clip_vision','sig_lora','unknown'}
assert record.get('type') in KNOWN_TYPES, f"Unknown model type: {record.get('type')!r}"

Type guard

def has_known_type(rec: dict) -> bool:
    return rec.get('type') in {'main','vae','controlnet','embedding','lora','t2i_adapter','ip_adapter','clip_vision','sig_lora'}

Try / catch

try:
    records.update_model(key, changes)
except ValidationError as e:
    if 'No model config class found' in str(e):
        log.error('Record %s has an unrecognized type; skipping or re-adding', key)
        records.del_model(key)
    else:
        raise

Prevention

When it happens

Trigger: Updating a model record (update_model) whose stored type value is not one of the known model types (e.g. a corrupted DB row, a type from a newer/older InvokeAI version, or hand-edited database).

Common situations: Database written by a different InvokeAI version where the type enum changed; manual SQL edits introducing a bad type string; plugin/custom model types not registered in the running process.

Related errors

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/9187f48dfa36fe19. Report an issue: GitHub.