Endemic: Local On-Device AI

Endemic runs large language models directly on your iPhone, iPad, and Mac. No API keys or cloud inference. Models and conversations stay on your device unless you choose to sync conversations through iCloud.

Choose a model from the curated catalog, download it, and chat without an internet connection. Endemic filters the catalog by available memory, so you only see models suited to your hardware.

Endemic includes Qwen 3.5 and Google's Gemma 4, from 0.8B models for recent iPhones to 9B models for Macs and iPad Pros with at least 16GB of memory. Switch models whenever you like.

QWEN 3.5 AND GEMMA 4, ON DEVICE

Qwen 3.5 models are included free, from the compact 0.8B that runs on any supported device to the more capable 9B for devices with at least 16GB of memory. Google's Gemma 4 supports full chat templating and function calling in Endemic. The 4B Gemma works well on a modern phone, while the 9B is better suited to iPad Pro and Mac. Both families support Local Web Search and Local Web Browser for current information.

Gemma 4 is available with Endemic Pro, a small monthly subscription that also includes conversation folders and supports ongoing development by one person. Qwen 3.5 is included free. You can try Pro and cancel any time.

WEB SEARCH AND BROWSING, ON DEVICE

Endemic can search the web and read pages from a conversation. Ask for current information or a specific site, and the model can call local tools to fetch results and summarize them. Requests go directly from your device over your network connection. There is no proxy or Folding Sky server in the middle.

Enable Local Tools in Settings to add the searchWeb and openWebPageLocally tools. Calls and results appear inline, so you can see what was requested and returned.

IMAGES IN, ANSWERS OUT

Vision-capable local models can read images attached to a message. Ask one to describe a photo, transcribe a screenshot, or explain a diagram. The image stays on your device, and the analysis works offline.

WHAT LOCAL INFERENCE ACTUALLY MEANS

During a conversation, the model runs on your device's CPU and GPU. Prompts and responses are not sent to an inference server. The work happens on your hardware.

A 4B model on a phone is useful for many tasks, but it will not match a frontier model running on datacenter hardware. For the strongest responses, use cloud AI. For private, offline work, use Endemic.

MODELS AND DEVICES

Endemic detects your hardware and recommends the strongest model that fits. An iPhone with 8GB of memory runs the 4B comfortably. iPads and Macs with at least 16GB handle the 9B. The 0.8B runs quickly on any supported device.

Models are open-weight GGUF files downloaded from public hosting and stored locally. Endemic excludes them from iCloud backup, so they do not use your iCloud storage quota.

PRIVACY

Endemic has no Folding Sky backend. Optional conversation sync uses your personal iCloud account.

Built by one person in Reno, Nevada.

Privacy Policy: https://folding-sky.com/privacy

Terms of Service: https://folding-sky.com/terms