vercel/ai · error · AISDKError

XAI_VIDEO_GENERATION_ERROR

XAI_VIDEO_GENERATION_ERROR

Error message

No request_id returned from xAI API.

What it means

In doStart of the xAI video model, the create-video API response must contain a request_id used to poll generation status. If the response body has no request_id (empty, malformed, or an error payload parsed as success), the model throws AISDKError named XAI_VIDEO_GENERATION_ERROR because generation cannot be tracked without it.

Source

Thrown at packages/xai/src/xai-video-model.ts:527

    } else {
      endpoint = `${baseURL}/videos/generations`;
    }

    const { value: createResponse, responseHeaders } = await postJsonToApi({
      url: endpoint,
      headers: combineHeaders(this.config.headers(), options.headers),
      body,
      failedResponseHandler: xaiFailedResponseHandler,
      successfulResponseHandler: createJsonResponseHandler(
        xaiCreateVideoResponseSchema,
      ),
      abortSignal: options.abortSignal,
      fetch: this.config.fetch,
    });

    const requestId = createResponse.request_id;
    if (!requestId) {
      throw new AISDKError({
        name: 'XAI_VIDEO_GENERATION_ERROR',
        message: `No request_id returned from xAI API.`,
      });
    }

    return {
      operation: { requestId },
      warnings,
      response: {
        timestamp: currentDate,
        modelId: this.modelId,
        headers: responseHeaders,
      },
    };
  }

  async doStatus(
    options: Parameters<NonNullable<VideoModelV4['doStatus']>>[0],

View on GitHub (pinned to 69428b1f8b)

Solutions

1. Retry the generation request; if it reproduces, inspect the raw response body/headers for an underlying error message.
2. Verify the API key has xAI video-generation access and the model id is valid.
3. Check for network middlewares (proxies, mocks, older fetch interceptors) that might alter the response body.
4. Update @ai-sdk/xai to the latest version in case the response handling was adapted to an API change.

function hasRequestId(r: unknown): r is { request_id: string } {
  return typeof (r as any)?.request_id === 'string' && (r as any).request_id.length > 0;
}

try {
  await videoModel.doStart({ prompt, ... });
} catch (e) {
  if (AISDKError.isInstance(e) && e.name === 'XAI_VIDEO_GENERATION_ERROR') {
    // log raw response context, retry with backoff; surface to user if persistent
  } else throw e;
}

When it happens

Trigger: POSTing a video generation request to xAI where the HTTP response succeeds (or is misparsed) but lacks request_id — e.g. API schema change, proxy stripping the body, auth/entitlement error returned as non-standard payload.

Common situations: Corporate proxy or gateway rewriting the response; xAI API contract change; account without video-generation access returning an unexpected body; transient API incidents.

Related errors

• FAL_VIDEO_GENERATION_ERROR
• xAI video status response exceeded ${MAX_PENDING_BODY_BYTES}
• ALIBABA_VIDEO_GENERATION_ERROR
• BLACK_FOREST_LABS_VIDEO_GENERATION_TIMEOUT
• BYTEDANCE_VIDEO_GENERATION_ERROR

AI-assisted analysis of vercel/ai@69428b1f8b (2026-08-30). Data as JSON: /api/errors/3d5501bd5c1e145b. Report an issue: GitHub.