Image Workbench
Recover image results from the last 24 hours, download originals in bulk, and compare personal and site trends.
Image Workbench recovers and organizes submitted image requests. Image Studio starts new generations and edits. Sign in with the same 1Route account; you do not need the original job ID or idempotency key.
You need an available account on the 1Route main site to use the workbench. If your first sign-in shows an unbound account, follow the main-site link to create an account or complete sign-in and binding, then return and choose “Check account again”.
Recover Images After a Disconnect
- Sign in to Image Workbench and open History.
- Filter by status, model, or prompt and open the task details.
- Download originals from successful tasks. A client that missed the response does not mean the server failed to finish.
- Select completed tasks to download several results in one ZIP.
The ZIP contains every final output original and a manifest with prompts, models, completion times, and filenames. Reference images, masks, and thought images are excluded. Select up to 100 tasks at a time.
The workbench displays image requests submitted through this site's unified image service. Refresh later for tasks still running. It does not resubmit or retry generation.
The history table shows complete group names, the model and protocol, requested and actual dimensions, and the stage and duration. Blue capsules indicate requests or running tasks; green indicates actual results or success. Running images use a dashed placeholder. Prompts show two lines; open the details for the full text.
Scroll down to load earlier recent records automatically, or choose “Load more”. Images load when they approach the visible area, and previews you already opened can be reused. Lists and details refresh manually: choose “Refresh records” or “Refresh details” for the latest status. Each account can refresh once every 5 seconds; rapid clicks are combined. Loading more records is unaffected by this interval.
Retention and Privacy
Only you can access your history. We maintain comprehensive encryption protection and automatic data cleanup standards for images and creative records.
The workbench displays images and history for 24 hours from request creation. After that window, they are cleared from the workbench and excluded from search, downloads, and statistics. Download images you wish to keep before that time.
Status Meanings
| Color | Task status | Key status |
|---|---|---|
| Green | Successful; results can be downloaded | Available |
| Blue | Running | — |
| Yellow | Queued | Quota or restrictions need attention |
| Red | Failed | Disabled, expired, or unavailable |
| Gray | Cancelled | — |
Statuses include text and icons. Expired content is cleared instead of becoming a browsable task status.
Compare Recent Trends
Overview compares personal and site success rates over the last 24 hours. Choose the same model for both curves. Personal task counts, image counts, and success rates are exact. Site percentages are approximate, rounded to five percentage points.
Success rate is successful tasks divided by successful plus failed tasks. Queued, running, and cancelled tasks are excluded. Hours without completed requests remain empty. Error-category shares use failed tasks as their denominator; they are not generation failure rates.
Suggestions use recorded final failure categories. They do not guess your network conditions or prompt quality. For parameter, content-policy, quota, timeout, or delivery issues, follow the corresponding advice to the FAQ or model guide.
Image Groups, Models, and Keys
Groups and Models includes GPT Image's general, official-Key, Adobe, mixed, and Microsoft Azure groups, plus Nano Banana. All group descriptions are visible; operations depend on your current account permissions.
Group names match the main site. The channel label in parentheses is part of the name:
| Group name | Channel description |
|---|---|
| GPT Image 🎨 | Image group using OpenAI Images calls |
| GPT Image 🧑🎨 (官Key渠道) | Official API Key channel |
| GPT Image 🖌️ (Adobe 渠道) | Adobe source channel |
| GPT Image 🌌 (神秘混合渠道) | Mixed source channel |
| GPT Image ☁️ (微软Azure渠道) | Microsoft Azure channel |
| Nano Banana 🍌 | Image model group using Gemini Native calls |
Click View models on an available group to open a dialog. Copy model IDs, read the protocol usage guidance, and open calling examples and documentation. The list shows base models and omits 1K/2K/4K size aliases and Nano Banana preview models; select image size through protocol parameters.
Opening the model dialog first reuses an existing available key for that group. If you have no key for the group, the workbench creates an Image Workbench default key and queries the models. If existing keys are unavailable, manage them on the main site; the workbench does not create another key to bypass restrictions.
API Keys supports viewing, copying, and creating keys. Creation asks only for a name and image group, using main-site defaults for other settings. Manage disabling, deletion, quotas, expiration, rate limits, and IP rules on the main site.
A model list shows what your key can currently read. It is not a health check. See protocol selection and model comparison when choosing how to call the service.