invoke-ai/InvokeAI · error · TypeError

Expected AutoencoderKL or FluxAutoEncoder for Z-Image VAE, g

Error message

Expected AutoencoderKL or FluxAutoEncoder for Z-Image VAE, got {type(vae_info.model).__name__}. Ensure you are using a compatible VAE model.

What it means

Z-Image image-to-latents encodes images with a VAE that must be an AutoencoderKL or FluxAutoEncoder instance. vae_encode is a static method that validates the loaded model type before encoding; anything else (wrong architecture's VAE) raises TypeError so encoding proceeds only with a compatible VAE.

Source

Thrown at invokeai/app/invocations/z_image_image_to_latents.py:46

@invocation(
    "z_image_i2l",
    title="Image to Latents - Z-Image",
    tags=["image", "latents", "vae", "i2l", "z-image"],
    category="latents",
    version="1.1.0",
    classification=Classification.Prototype,
)
class ZImageImageToLatentsInvocation(BaseInvocation, WithMetadata, WithBoard):
    """Generates latents from an image using Z-Image VAE (supports both Diffusers and FLUX VAE)."""

    image: ImageField = InputField(description="The image to encode.")
    vae: VAEField = InputField(description=FieldDescriptions.vae, input=Input.Connection)

    @staticmethod
    def vae_encode(vae_info: LoadedModel, image_tensor: torch.Tensor) -> torch.Tensor:
        if not isinstance(vae_info.model, (AutoencoderKL, FluxAutoEncoder)):
            raise TypeError(
                f"Expected AutoencoderKL or FluxAutoEncoder for Z-Image VAE, got {type(vae_info.model).__name__}. "
                "Ensure you are using a compatible VAE model."
            )

        # Estimate working memory needed for VAE encode
        estimated_working_memory = estimate_vae_working_memory_flux(
            operation="encode",
            image_tensor=image_tensor,
            vae=vae_info.model,
        )

        with vae_info.model_on_device(working_mem_bytes=estimated_working_memory) as (_, vae):
            if not isinstance(vae, (AutoencoderKL, FluxAutoEncoder)):
                raise TypeError(
                    f"Expected AutoencoderKL or FluxAutoEncoder, got {type(vae).__name__}. "
                    "VAE model type changed unexpectedly after loading."
                )

View on GitHub (pinned to 0b6a024f2f)

Solutions

1. Connect a Z-Image-compatible VAE (AutoencoderKL or FluxAutoEncoder) to the invocation's 'vae' input
2. Check the model manager that the VAE entry points at the correct architecture's file
3. Replace the VAE node in the workflow with one matching the Z-Image base model

Example fix

// before
img2latents = ZImageImageToLatentsInvocation(image=img, vae=sdxl_vae_field)
// after
img2latents = ZImageImageToLatentsInvocation(image=img, vae=z_image_vae_field)

vae_info = context.models.load(vae_field.vae)
if not isinstance(vae_info.model, (AutoencoderKL, FluxAutoEncoder)):
    raise TypeError(f"Incompatible VAE: {type(vae_info.model).__name__}")

def is_zimage_vae(model) -> bool:
    return isinstance(model, (AutoencoderKL, FluxAutoEncoder))

try:
    latents = img2latents.invoke(context)
except TypeError as e:
    if "Expected AutoencoderKL or FluxAutoEncoder for Z-Image VAE" in str(e):
        raise ModelCompatibilityError("swap in a Z-Image-compatible VAE") from e
    raise

When it happens

Trigger: Calling invoke() on the Z-Image image-to-latents invocation where the VAE referenced by the VAEField loads to a model that is neither AutoencoderKL nor FluxAutoEncoder — e.g. an SDXL/SD-1 VAE or other architecture's autoencoder.

Common situations: Wiring a VAE from a different model family into a Z-Image workflow; a model-manager entry pointing at the wrong VAE file; using an old workflow whose VAE key refers to a now-different model.

Related errors

• Expected AutoencoderKLQwenImage or AutoencoderKLWan, got {ty
• Expected AutoencoderKLWan or FluxAutoEncoder for Anima VAE,
• Expected AutoencoderKLWan or FluxAutoEncoder, got {type(vae)
• Expected AutoencoderKLWan or FluxAutoEncoder for Anima VAE,
• Expected AutoencoderKLWan or FluxAutoEncoder, got {type(vae)

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/702fb24a1560d6d2. Report an issue: GitHub.