n8n-io/n8n · error · Error

No image data returned from Gemini API

Error message

No image data returned from Gemini API

What it means

Thrown by the GoogleGemini image edit operation when the API response passes the shape check (has candidates with content.parts) but no part contains inlineData with actual base64 image data. This means the model processed the request but did not produce an image output — the candidate parts contain only text or other non-image content.

Source

Thrown at packages/@n8n/nodes-langchain/nodes/vendors/GoogleGemini/actions/image/edit.operation.ts:215

	const response: unknown = await apiRequest.call(
		this,
		'POST',
		`/v1beta/${model}:generateContent`,
		{
			body,
		},
	);

	if (!isGenerateContentResponse(response)) {
		throw new Error('Invalid response format from Gemini API');
	}

	const promises = response.candidates.map(async (candidate) => {
		const imagePart = candidate.content.parts.find((part) => 'inlineData' in part);

		// Check if imagePart exists and has inlineData with actual data
		if (!imagePart?.inlineData?.data) {
			throw new Error('No image data returned from Gemini API');
		}

		const mimeType = imagePart.inlineData.mimeType;
		const fileName = getFilenameFromMimeType(mimeType, 'image', 'png');
		const bufferOut = Buffer.from(imagePart.inlineData.data, 'base64');
		const binaryOut = await this.helpers.prepareBinaryData(bufferOut, fileName, mimeType);
		return {
			binary: {
				[outputKey]: binaryOut,
			},
			json: {
				...binaryOut,
				data: undefined,
			},
			pairedItem: { item: i },
		};
	});

View on GitHub (pinned to 5ac6606e81)

Solutions

1. Verify the model ID supports image output — the default fallback is 'models/gemini-2.5-flash-image-preview', but an explicitly selected model may be text-only.
2. Inspect the response parts — the model may have returned text explaining the issue. Log candidate.content.parts to see what the model actually said.
3. Refine the edit prompt to be a clear, actionable image editing instruction.
4. If the content is being safety-filtered, adjust the prompt or input images to comply with Gemini's content policies.

function hasImageData(candidate: { content: { parts: Array<Record<string, unknown>> } }): boolean {
  const imagePart = candidate.content.parts.find(part => 'inlineData' in part);
  return Boolean(imagePart && (imagePart as { inlineData?: { data?: string } }).inlineData?.data);
}

try {
  // ... process Gemini response ...
} catch (error) {
  if (error instanceof Error && error.message === 'No image data returned from Gemini API') {
    // The model may have returned text instead of an image — check candidate parts
    console.warn('No image data. Model may have returned a text response or refusal.');
    // retry with a clearer prompt or different model
  }
  throw error;
}

When it happens

Trigger: The Gemini model returns a response with valid candidates but the parts array has no element with 'inlineData' or the inlineData.data field is empty/falsy. This occurs when the model responds with text instead of an image (e.g. explaining why it can't edit the image), when the model returns a refusal, or when the model version doesn't support image output modalities.

Common situations: Using a text-only Gemini model instead of an image-capable model (gemini-2.5-flash-image-preview); the edit prompt is ambiguous and the model responds with text clarification instead of an image; content safety filter causes the model to refuse image generation; API issue where the image data is stripped from the response.

Related errors

• Invalid response format from Gemini API
• A non-empty prompt is required.
• Invalid images parameter format
• There was an error: "The workflow did not return a response"
• There was an error: "The workflow did not return a response"

AI-assisted analysis of n8n-io/n8n@5ac6606e81 (2026-08-12). Data as JSON: /api/errors/e16906a0cfe7eeb6. Report an issue: GitHub.