> ## Documentation Index > Fetch the complete documentation index at: https://docs.acedata.cloud/llms.txt > Use this file to discover all available pages before exploring further. # Use GPT Image in n8n > Install the independent GPT Image node, save an Ace Data Cloud API key, and retrieve an existing task. **Ace Data Cloud** maintains this integration for its GPT Image API; it is not an official node from the model developer. It supports image generation and editing and asynchronous task retrieval. A direct model-provider subscription is not an Ace Data Cloud credential. The npm package is published, but it was absent from the production verified-node catalog read on 2026-10-19. npm publication does not establish n8n Cloud verification. This page describes the new public npm release from 2026-10-19, including service-name aliases and native credential help. npm publication does not automatically update Cloud verification. Package documentation remains English to meet n8n verification rules; Chinese customer instructions live separately. ## Install and find the node 1. On self-hosted n8n, an owner/admin follows [n8n's installation guide](https://docs.n8n.io/integrations/community-nodes/installation-and-management/gui-installation/), opens **Settings → Community nodes → Install**, and enters `@acedatacloud/n8n-nodes-gpt-image@0.1.2`. Install only in an instance you trust and administer. 2. On Cloud, use the [verified-node picker](https://docs.n8n.io/integrations/community-nodes/installation-and-management/install-verified-community-nodes/). If absent, do not change security settings or create a duplicate listing. 3. Add **Manual Trigger**, or use your existing trigger. Select its next **+**, search **GPT Image**, and choose **GPT Image by Ace Data Cloud**; older versions use **by AceDataCloud**. Workflow node identifiers stay the same. This is an action/query node, not a trigger. ![Official n8n reference: add an action after a trigger](https://raw.githubusercontent.com/n8n-io/n8n-docs/4a84a4229f694c8565dcd0a9cd0af37623fa2cbc/docs/get-started/.gitbook/assets/build-your-first-workflow-get-weather-01.png) Unchanged official tutorial image, referenced 2026-10-18; capture date and exact version are not declared. It shows HTTP Request search layout, **not this service's installation or output**. [Official source](https://docs.n8n.io/get-started/build-your-first-workflow). ## Register and prepare a dedicated API key 1. Open [Ace Data Cloud Applications](https://platform.acedata.cloud/console/applications?utm_source=opensource\&utm_medium=referral\&utm_campaign=evergreen\&utm_content=n8n_gpt_image_en_api_key), register or sign in, and choose **Initialize General Application** if **General application** is missing. 2. Copy its **API Key**, or use **Keys / Manage Keys → Create** for a separately named n8n key. General keys use services available to your account; dedicated applications are limited to their service. 3. Check **Expiration Time**, **Set Usage Limitation**, **Allowed APIs**, GPT Image model/service access, balance and [current pricing](https://platform.acedata.cloud/models?utm_source=opensource\&utm_medium=referral\&utm_campaign=evergreen\&utm_content=n8n_gpt_image_en_pricing). Task-read access does not grant creation permission; any trial allowance is whatever your account displays now. ![Real platform reference: copy a key or open Manage Keys](https://raw.githubusercontent.com/AceDataCloud/SunoDify/8d3d03570e73324220f111f7fe343da0792a8071/_assets/tutorial/get-api-key-en.png) ![Real platform reference: create a named key and review restrictions](https://raw.githubusercontent.com/AceDataCloud/SunoDify/8d3d03570e73324220f111f7fe343da0792a8071/_assets/tutorial/create-api-key-en.png) These are real platform references pinned to a source commit, referenced 2026-10-18. They are not n8n execution screenshots or proof of your account's access/balance. Keep real tokens out of prompts, screenshots and shared workflows. ## Save the native credential Open the GPT Image node and create **GPT Image by Ace Data Cloud API**; older versions use **by AceDataCloud API**. Paste only the raw token into **API Token**, without `Bearer `, quotation marks or spaces. A `platform-...` account-management token cannot authorize generation/task retrieval. Save and select this credential on each node that needs access. The current credential test queries an empty task batch and does not create media; passing it does not establish model entitlement or access to a specific task. The key stays in n8n credential storage and is omitted from the downloaded example. ## Retrieve your existing task first 1. Download the [query-only workflow](https://raw.githubusercontent.com/AceDataCloud/Docs/main/examples/n8n/read-gpt-image-task.json) and choose **Import from URL/File**. It starts inactive and contains a trigger, note and GPT Image task reader, with no generation or model node. 2. Choose **Resource → Task**, **Operation → Get**, select your credential, and replace **Task ID** with an existing ID from this service that you may read. An ID is not a prompt or media URL. Do not execute the placeholder. 3. Run that reader once. Continue only when `finished` and `successful` are true; final media links are in `data`. For processing tasks, retrieve the **same ID** later. Preserve failures/cancellations/auth errors instead of creating a replacement task. Package/import/missing-credential validation does not establish an authorized service response. Validate ordinary installation, credential selection and output in your instance. Querying an old task is not a new generation or billing receipt. ## Create once and wait for the result After checking the read path, use the [source package's creation/polling examples](https://github.com/AceDataCloud/GPTImageN8N/tree/main/examples). Select **Resource → Image**, **Operation → Generate** and a matching model. `gpt-image-2` is an existing example value; check current entitlement and price before running. Each input item may create a task. Check budget before batching. Run creation once and leave **Retry On Fail** off. Save `taskId`, poll the same ID with a delay, and bound waiting with explicit timeout/failure handling. An acknowledgment is not final media, and a workflow timeout does not cancel a server task. Inspect the original ID and usage before manually retrying an ambiguous request. New creation requires a separate budget; this guide has not submitted a paid request for you. Consumption is in Credits, not USD. Use the matching single usage record, not token counts or example duration, to establish the actual charge. ## Troubleshooting | Symptom | Action | | - | - | | Node missing | Distinguish verified Cloud discovery from self-hosted npm installation; check the exact package/version and search the action panel after a trigger. | | 401 / 403 | Check raw token, service/model/task access, expiry and Allowed APIs. Stop; do not bypass identity or create a replacement generation. | | 429 | Reduce concurrency and follow the response delay; do not replay a possibly accepted creation. | | Creation timeout / 5xx | Inspect the original execution, task ID and usage before retrying; prefer retrieving that ID. | | Task not finished | Inspect `finished`, `successful` and `error`; HTTP 200 or an outer success flag alone is not completion. | [Source and support](https://github.com/AceDataCloud/GPTImageN8N) · [Back to n8n overview](/en/integrations/n8n). Include package/n8n version, operation and a redacted error when reporting an issue. Omit API keys and private inputs. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.