Offline AI Studio

Offline AI Studio brings private, offline-first AI to iPhone and iPad.

Chat with open-weight language models running on your device, generate images locally with Core ML, transcribe speech with Whisper, extract text with OCR, and use focused AI tools without relying on cloud AI by default.

Why people use Offline AI Studio
• Local by default: with on-device models selected, chat, image generation, Whisper transcription, and OCR stay on your device
• Works offline after model download: useful while traveling, on flights, or in restricted networks
• Optional cloud mode on your terms: connect your own API key for supported providers or custom endpoints
• One-time purchase with optional lifetime Pro unlock. No subscription.

What you can do
• Chat with on-device AI models
• Generate images locally with Core ML Stable Diffusion
• Use on-device Whisper speech-to-text
• Extract text from photos, scans, and files with OCR
• Use AI Tools for summarizing, rewriting, grammar, translation, brainstorming, document workflows, and more
• Experiment with Code Lab, a lightweight prototype sandbox for HTML, CSS, and JavaScript
• Manage local GGUF models, import compatible GGUF files, and tune advanced settings

Offline AI Studio is built for people who want more privacy and more control.
Keep everything local for maximum privacy, or enable API mode with your own key when you want access to larger cloud models.

Good to know
• Offline AI Studio is not a replacement for large cloud AI systems
• Speed and quality depend heavily on your hardware
• Larger models need more RAM, storage, and patience
• Best experience is on newer devices
• Internet is still used for model downloads, optional web search, and any API requests you enable

Model note
Offline AI Studio runs supported open-weight models locally on-device and includes in-app license references for the currently offered model families and image weights. Current on-device text model families include LFM, Llama, Qwen, Gemma, and SmolLM. On-device image generation uses Core ML Stable Diffusion tiers including SD 1.5, SD 2.1, and SDXL.