Skip to main content

AI Models

Download AI models to run script generation, TTS, and image creation directly in your browser.

Llama 3.2 3B

High-quality script generation, content writing, translations

AI WriterScript GenerationRemix Hub
1.9 GB4.0 GB VRAMLLM
Requires high-end GPU — may be slow on this device

Llama 3.2 1B

Good quality scripts and writing, lower GPU requirements

AI WriterScript GenerationRemix Hub
763 MB2.0 GB VRAMLLM
Requires medium GPU — may be slow on this device

Qwen 0.5B

Lightweight LLM for basic content generation

AI Writer (basic)
381 MB1.0 GB VRAMLLM
Requires low GPU — may be slow on this device

Kokoro TTS

Text-to-speech with 54+ voices and viseme lip-sync data

Voice StudioVideo TTSAvatar Lip-Sync
115 MB200 MB VRAMTTS
Requires low GPU — may be slow on this device

Whisper Tiny

Fast speech-to-text for auto-captions and video translation

Auto-CaptionsVideo Translation
42 MB100 MB VRAMSTT
Requires low GPU — may be slow on this device

Whisper Base

Higher accuracy speech-to-text for captions and transcription

Auto-Captions (better)
76 MB200 MB VRAMSTT
Requires medium GPU — may be slow on this device

Model Storage

Model folder

Models are stored in browser storage on this device.

Your browser does not support choosing a real folder (File System Access API) — models are kept in browser storage. Chrome and Edge support folders.

Llama 3.2 3B1.9 GBBrowser (engine-managed)not_downloaded
Llama 3.2 1B763 MBBrowser (engine-managed)not_downloaded
Qwen 0.5B381 MBBrowser (engine-managed)not_downloaded
Kokoro TTS115 MBBrowsernot_downloaded
Whisper Tiny42 MBBrowsernot_downloaded
Whisper Base76 MBBrowsernot_downloaded

Site storage

Storage usage unavailable in this browser.

Orphaned data

No orphaned model data found.

Clear all app data

Wipes this site's caches, databases and private file storage. Models in your chosen folder are kept and re-detected.

Advanced Options

Download Source

Models download directly from the HuggingFace CDN to your device and are cached for offline use. Nothing is uploaded and no data passes through Contentine servers.

Clear Non-Model Cache

Remove cached generated images and audio. Model weights are kept.