spring-projects/spring-ai · error · IllegalArgumentException

Unsupported media type:

Error message

Unsupported media type: 

What it means

If a Media attached to the document is neither an accepted image, nor video, nor text mime type, the model throws IllegalArgumentException. This is the final else branch of the media dispatch in doSingleDocumentPrediction and indicates input content the Vertex AI multimodal embedding API cannot embed.

Source

Thrown at models/spring-ai-vertex-ai-embedding/src/main/java/org/springframework/ai/vertexai/embedding/multimodal/VertexAiMultimodalEmbeddingModel.java:211

					logger.warn("Unsupported image mime type: " + media.getMimeType());
					throw new IllegalArgumentException("Unsupported image mime type: " + media.getMimeType());
				}
			}
			else if (media.getMimeType().isCompatibleWith(VIDEO_MIME_TYPE)) {
				instanceBuilder.video(VideoBuilder.of(media.getMimeType())
					.videoData(media.getData())
					.startOffsetSec(mergedOptions.getVideoStartOffsetSec())
					.endOffsetSec(mergedOptions.getVideoEndOffsetSec())
					.intervalSec(mergedOptions.getVideoIntervalSec())
					.build());
				documentMetadata.put(ModalityType.VIDEO,
						new DocumentMetadata(document.getId(), media.getMimeType(), media.getData()));
			}
			else {
				if (logger.isWarnEnabled()) {
					logger.warn("Unsupported media type: " + media.getMimeType());
				}
				throw new IllegalArgumentException("Unsupported media type: " + media.getMimeType());
			}
		}

		List<Value> instances = List.of(VertexAiEmbeddingUtils.valueOf(instanceBuilder.build()));

		PredictRequest.Builder predictRequestBuilder = PredictRequest.newBuilder()
			.setEndpoint(endpointName.toString())
			.setParameters(VertexAiEmbeddingUtils.jsonToValue(jsonHelper.toJson(Map.of())))
			.addAllInstances(instances);

		PredictResponse embeddingResponse = client.predict(predictRequestBuilder.build());

		int index = 0;
		List<Embedding> embeddingList = new ArrayList<>();
		for (Value prediction : embeddingResponse.getPredictionsList()) {
			if (prediction.getStructValue().containsFields("textEmbedding")) {
				Value textEmbedding = prediction.getStructValue().getFieldsOrThrow("textEmbedding");
				float[] textVector = VertexAiEmbeddingUtils.toVector(textEmbedding);

View on GitHub (pinned to 98a7beda4f)

Solutions

1. Remove or replace the unsupported media with image, video, or text content
2. Use the correct Spring AI model for other content (e.g. a text embedding model for text-only docs)
3. Explicitly set the MimeType so it matches one of the supported types
4. Pre-validate Media mime types in your ingestion pipeline and skip/log unsupported items

Example fix

// before
doc.getMedia().add(new Media(MimeTypeUtils.parseMimeType("audio/mpeg"), audioBytes)); // throws
// after
Media media = new Media(MimeTypeUtils.parseMimeType("audio/mpeg"), audioBytes);
if (!isSupportedMime(media.getMimeType())) { log.warn("skipping unsupported media"); return; }

boolean embeddable = doc.getMedia().stream().allMatch(m -> m.getMimeType().getType().equals("image") || m.getMimeType().getType().equals("video") || m.getMimeType().getType().equals("text"));

static boolean isEmbeddableMedia(MimeType mt) { return List.of("image","video","text").contains(mt.getType()); }

try { response = model.call(request); } catch (IllegalArgumentException e) { if (e.getMessage().startsWith("Unsupported media type")) { /* split or drop the doc */ } throw e; }

When it happens

Trigger: Building a Document whose Media MimeType is incompatible with IMAGE_MIME_TYPE, VIDEO_MIME_TYPE and TEXT_MIME_TYPE (e.g. audio/audio-mpeg, application/pdf-as-binary when not routed to document handling, application/zip) and calling the model.

Common situations: Trying to embed audio clips with the multimodal embedding model; passing arbitrary binary files with application/octet-stream; forgetting that the API supports only image/video/text modalities.

Understand the failure class

Background: UnsupportedOperationException and "is not supported" errors: when a library deliberately refuses a call — this error's family across 30 libraries.

Related errors

• Unsupported image mime type:
• model cannot be null or empty
• java.io.IOException
• Title is only valid with task_type=RETRIEVAL_DOCUMENT
• Method must have exactly 1 parameter (List<McpSchema.Resourc

AI-assisted analysis of spring-projects/spring-ai@98a7beda4f (2026-09-11). Data as JSON: /api/errors/c5f34477714a5ea8. Report an issue: GitHub.