FAQ
Everything you need to know about SilicaAI. Use our Support page if your question isn't answered here.
What is SilicaAI?
Silica is a local-first macOS AI productivity application. It integrates multi-turn chatting, voice-to-text, agent workflows, presentations, and image generation. All core tools run natively on your Apple Silicon Mac for maximum privacy and performance.
Does it work offline?
Yes, entirely. The primary language models, Whisper transcription, and image generation are downloaded strictly to your Mac. Internet is only used if you explicitly enable features like web-search agents or cloud fallbacks.
Does it send my prompts to the cloud?
No. SilicaAI was built so your data never has to leave your machine unless you specifically enable third-party cloud integrations inside the settings.
What Macs are supported?
SilicaAI requires macOS 14+ (Sonoma or higher). While it functions on Intel Macs, we strongly recommend an Apple Silicon Mac (M1/M2/M3/M4) with 16GB of unified memory for the best LLM performance.
Do I need to install models myself?
No, SilicaAI detects your hardware and auto-recommends optimized model downloads. You can also manually import GGUF and MLX models if you are a developer.
Is this completely free?
Yes. SilicaAI is free to download and use. We have a donation model for users who find value in the tool and wish to support its ongoing development.
Why is it in Beta?
Because autonomous agents, local image diffusion, and voice dictation are highly dependent on rapidly evolving open-source models, some features may feel experimental. The Beta tag remains until the core platform is rock solid.