Does the enhancer call an AI model?

No. Enhancement is local template rewriting only. You paste the result into your own LLM; we never see prompts.

Which models do briefs target?

Model-agnostic markdown briefs work in GPT-5.6, Claude 5, Gemini 3, DeepSeek V4, Grok 4.5, and local models.

Only optionally in localStorage history on your device. Nothing syncs to servers.

Count prompt tokens, estimate context usage, and compare costs across GPT-5.6, Claude 5, Gemini 3, DeepSeek V4, and Grok 4.5.

Disrupt SynthID watermarks, purge C2PA content credentials, and strip generative AI metadata in-browser.

Inputs stay in browser memory. No external request is required for this operation, and zero data leaves your device.