Short answer: there isn’t a consumer “incognito/temporary” mode that stops Text-to-Speech generations from appearing in History. By default, generations are preserved in your account history, but you can delete them at any time.
What you can do today
Delete items right after you generate and download them:
In the web app, your generations appear under History, where you can remove them. (help.elevenlabs.io)
Via API, list your history items and delete the ones you don’t want to keep:
GET /v1/history, then DELETE /v1/history/:history_item_id. (elevenlabs.io)
Opt out of having your content used to improve models: Profile icon > Terms and privacy > Data use > disable “Improve the models for everyone.” This does not hide History, but it stops new submissions from being used for training. (help.elevenlabs.io)
Your cloned voices remain private to your account unless you explicitly share them. (help.elevenlabs.io)
If you need true “no storage”
Enterprise-only: Zero Retention Mode (ZRM) ensures request text/audio aren’t logged or kept (no items appear in History). It works for API-based TTS/STS by sending enable_logging=false, and can also be enforced per-Agent or workspace-wide. (elevenlabs.io)
For Conversational Agents, you can set data retention down to 0 days (immediate deletion). (elevenlabs.io)
Note: outside ZRM, deleted items are removed from the database immediately, but backups may retain them for up to 30 days before expiring. (elevenlabs.io)
Practical tip for personal messages
Generate, download, then promptly delete the History item; also disable model-improvement data use in settings. This is the most private workflow available on non‑Enterprise plans today. (elevenlabs.io)
If you need ZRM, contact Sales about Enterprise access. (elevenlabs.io)
hidden gorge
knotty zinc