Blog /
Chrome Nano vs WebLLM vs Gemini cloud
Compare Quick Rephrase engines: on-device Nano, optional WebLLM, and optional Gemini cloud — privacy, setup, and when to use each.
- engines
- nano
- webllm
- cloud
Quick Rephrase v2 can use three engines. Here’s how to choose.
On-device Nano (default)
| Best for | Short rewrites, privacy, no setup cost |
| Needs | Chrome with on-device AI; Install from Settings if prompted |
| Leaves device? | No |
WebLLM (optional)
| Best for | Local rewrites when Nano isn’t available |
| Needs | WebGPU + one-time ~1 GB model download |
| Leaves device? | No |
Gemini cloud (optional)
| Best for | Long text, reliability when local engines fail |
| Needs | Your Gemini API key |
| Leaves device? | Selected text only, when cloud runs |
Recommended defaults
- Keep primary = On-device (Nano).
- Install Nano from Settings if status isn’t Ready.
- Add WebLLM only if you want a local backup.
- Add a cloud key only if you want fallback or Cloud as primary.
Session tip: the in-page popover has a provider dropdown for that rewrite only — Settings primary stays unchanged.