Venice treats failed generations and unsatisfactory completed outputs differently.
Failed generations (error, no usable output, session expired mid-run, unexpected error, or a run lost while queued/processing) should be automatically refunded when applicable. You can review credit usage and refunds by opening Settings from the left-hand menu and going to Usage. That page shows whether a generation failed and whether it was refunded. After a balance change, refresh the page and restart Venice to see the updated balance.
If any of the failures did not get refunded, contact support with a screenshot from Settings → Usage of any failed generations and we will make it right. Please also send the Task ID, which can be copied by clicking the 3 dots next to any generation.
Additionally, consider batching multiple Task IDs and error notes before contacting support to minimize workflow interruptions. The Usage screen also displays the amount charged for each generation, which support can reconcile against the ledger to resolve discrepancies.
Credit reissues require a human agent to action, so having the Task ID and screenshot ready before you contact support avoids back-and-forth. Note that refunds cannot be issued for generations that have been deleted, since they can no longer be verified — keep the Task ID for any failed generation before deleting it.
If a generation completed but the result missed the look you wanted (anatomy, details, edit fidelity, prompt match, or similar), that is an output-quality issue rather than a failure. Once a generation is completed, the compute resources are consumed and the charge is considered final, so credits are not returned by default for completed results that missed the desired look. For those cases, improve the prompt or workflow instead:
Use Prompt Enhancer when you want help fleshing out a prompt with more detail. Pair it with clear constraints: what must stay the same, what must change, and what to avoid.
Turn Prompt Enhancer off when you need the model to follow a specific, exact prompt with no extra changes.
For stronger model-specific prompting, ask a web scraping model to research prompting techniques for that image or video model, then paste the refined prompt into the Image, Video, or Edit flow.
To reduce the chance of another lost run while a generation is processing:
Keep the Venice page or app open until the generation finishes or clearly fails.
Do not refresh, reload, or navigate away from the generation page while it is running.
Avoid rapid repeated Generate clicks after an error. Pause, wait for a clear failure or refund, then try once more.
If you are on mobile, keep the app or browser active and the device awake until it completes.
Best Practices for Data Persistence
To avoid losing access to completed generations after a browser data reset, download completed generations as soon as they are available. The downloaded file serves as a permanent copy, even if browser data is cleared.