Spell Checker
BetaReview spell with clear inputs, explained results, and relevant limitations.
Optional local AI · WebGPU requiredA simpler way to use the spell checker
Review spell with clear inputs, explained results, and relevant limitations. A small Qwen model generates an editable draft on your device. Output quality varies; verify facts and adapt the wording before use.
How it works
- Download and load the optional local model in a WebGPU-compatible browser.
- Describe your facts, audience, and purpose in up to 1,500 characters, then choose writing preferences.
- Generate a draft, review and edit it, then copy it. Reset clears your topic and output.
Your data stays with you.
Your prompts and drafts stay on this device. After you choose to load the model, files are downloaded from Hugging Face and the MLC project; those hosts receive normal download request information. Model files may be cached in your browser. Use Remove cached model to clear them.
What does Spell Checker do?
Review spell with clear inputs, explained results, and relevant limitations.
- Cost and account
- Free to use. No account required.
- Processing
- AI inference in this browser
- Device requirements
- A WebGPU-compatible browser and about 2 GB of available GPU memory. Mobile support depends on the device.
Frequently asked questions
Is this hosted AI?
No. Model files are downloaded only after you choose to load them. Prompts and drafts stay in this browser. The model needs about 2 GB of GPU memory.
Can I rely on the draft without checking it?
Review every draft. Small models can repeat text, miss instructions, and invent details. This beta tool does not verify facts, qualifications, or product claims.