Custom servers must support OpenAI-compatible API. Enable CORS in server
settings.
Checking WebGPU support...
GPU in use: Unknown
Runs entirely on your device - no API key and no server. The size shown for
each model is the approximate GPU memory it needs; the model downloads once and is cached by your
browser.Wrong GPU?
1. Windows: Settings → Display → Graphics → your browser → High
performance
2. NVIDIA: Control Panel → 3D Settings → Program Settings
3. or launch the browser with --force_high_performance_gpu
4. Restart browser and hit Re-check.
Keys are stored only for your session
A preset selects its own model and tool budget, so it takes
priority over the model chosen above.
Models are fetched from the selected provider after
validating your API key.
In-browser models are free and run entirely on your
device - prompts are never sent anywhere, and the download is cached for next time.
Visit any AI tool page to start using powerful AI-driven features like story generators, quiz
creators, and more.
Your API key is stored securely in your browser session and never sent to our servers.
Audio & Video Transcriber
Convert audio and video files to text using Whisper or Voxtral. If the file is larger, I reccomend transcribing locally since context length is limited.
⚠️ Unsupported Provider: Transcription requires OpenAI (Whisper) or Mistral (Voxtral). Please switch your provider in the header settings.