can1357/oh-my-pi · error · Error

xAI image edits accept up to ${XAI_MAX_EDIT_IMAGES} referenc

Error message

xAI image edits accept up to ${XAI_MAX_EDIT_IMAGES} reference images; got ${resolvedImages.length}.

What it means

xAI's image edit endpoint accepts at most XAI_MAX_EDIT_IMAGES reference images. When the edit request (input images present) exceeds that constant, the tool rejects the call before making any network request. It is a client-side pre-validation of the provider's documented limit.

Source

Thrown at packages/coding-agent/src/tools/image-gen.ts:1489

					if (provider === "xai") {
						if (!ctx.modelRegistry) {
							throw new Error("Missing modelRegistry for xAI image generation");
						}
						const xaiCreds = await resolveXAIHttpCredentials(ctx.modelRegistry, resolvedModel);
						if (!xaiCreds) {
							throw new Error(
								"No xAI credentials. Run /login → xAI Grok OAuth (SuperGrok or X Premium+) or set XAI_API_KEY.",
							);
						}

						const prompt = assemblePrompt(params);
						const aspectRatio = params.aspect_ratio ?? "1:1";
						const xaiResolution = resolveXAIResolution(params.image_size);

						const isEdit = resolvedImages.length > 0;
						if (isEdit && resolvedImages.length > XAI_MAX_EDIT_IMAGES) {
							throw new Error(
								`xAI image edits accept up to ${XAI_MAX_EDIT_IMAGES} reference images; got ${resolvedImages.length}.`,
							);
						}

						const xaiBaseBody: XAIImageRequestBase = {
							model: resolvedModel,
							prompt,
							aspect_ratio: aspectRatio,
							resolution: xaiResolution,
							n: 1,
							response_format: "b64_json",
						};
						const xaiBody: XAIImageRequestBody = isEdit
							? buildXAIEditPayload(xaiBaseBody, resolvedImages)
							: xaiBaseBody;
						const xaiEndpoint = isEdit ? "/images/edits" : "/images/generations";

						const xaiKey: ApiKey = ctx.modelRegistry.resolver(xaiCreds.provider, {

View on GitHub (pinned to 9690622007)

Solutions

  1. Reduce the number of input images to XAI_MAX_EDIT_IMAGES or fewer before retrying.
  2. Split the work into multiple sequential edit calls, each within the limit.
  3. Use a provider with a higher reference-image cap (e.g. gemini/openrouter) for this request.
  4. Keep only the most relevant reference images; providers weight the first images more anyway.

Example fix

// before
images: [img1, img2, img3, img4, img5, img6]
// after
images: [img1, img2] // within XAI_MAX_EDIT_IMAGES
Defensive patterns

Strategy: validation

Validate before calling

const XAI_MAX_EDIT_IMAGES = 10; // match provider limit
if (isEdit && images.length > XAI_MAX_EDIT_IMAGES) {
  images = images.slice(0, XAI_MAX_EDIT_IMAGES); // or split into batches
}

Type guard

const withinXaiEditLimit = (n: number, max: number): boolean => n > 0 && n <= max;

Try / catch

try {
  await imageGen({ images, ...params });
} catch (err) {
  if (err instanceof Error && /reference images/.test(err.message)) {
    // retry with first N images or split into multiple calls
  } else throw err;
}

Prevention

When it happens

Trigger: Calling image-gen with edit mode (one or more input images) where resolvedImages.length > XAI_MAX_EDIT_IMAGES, e.g. attaching too many reference images in a single request, at image-gen.ts:1489.

Common situations: Passing a whole folder of reference images to compose into one edit; piping multiple prior tool outputs as inputs; confusing per-request limits with per-conversation attachment limits.

Related errors

AI-assisted analysis of can1357/oh-my-pi@9690622007 (2026-08-31). Data as JSON: /api/errors/e73e0dd0afc83989. Report an issue: GitHub.