CoreLLM
CoreLLM
Private. Yours. Forever.
Did you know?
OpenAI
OpenAI retains chats under its policies: "temporary" conversations are deleted within about 30 days. But there's a catch — legal holds like ongoing litigation can force OpenAI to retain your data far longer than you'd expect (many of which are going on RIGHT NOW). In practice, you can't guarantee that what you delete is truly gone.
Anthropic
Anthropic now retains chats for up to 5 years if you allow training. If you opt out, your data is still kept for about 30 days for operational reasons. Even deleted chats may linger during that window — the timeline is never fully under your control.
Google, Meta, and others
Google's Gemini keeps conversations for up to 72 hours when history is off, and other providers vary. And we all know how much Meta LOVES to collect data about you. The truth is, all these companies integrate your activity across ecosystems, and you have little visibility into how long it persists or how it's used.
The Reality
Your data is not fully in your hands. Across today's AI platforms, your data is always someone else's asset. Data is their new currency, and you give up control while they keep the leverage.
Introducing CoreLLM
With CoreLLM, it's different: your chats never leave your device. No hidden retention windows, no training without consent, no excuses. Your conversations remain yours.
At its core, CoreLLM is a Progressive Web App (PWA). You can still use it in the browser similar to ChatGPT, or you can install it on your device. However, unlike ChatGPT, you will be able to use it offline, and all your conversations will be stored on your device. Keep your existing workflows, but with true privacy.
Some more features:
Free & Open Source
CoreLLM will be free and open source. No need to explain why this is important.
Same UI as ChatGPT
All the basic functionality of ChatGPT, with some extra features like more model selection.
Truly Private
All chats stay on your device. Any referenced files are also stored on your device. They can be exported anytime.
Your Choice of Models
Choose between DeepSeek, Qwen, Gemma, Meta Llama, and more. Switch models based on your specific needs and preferences.
Agentic Memory & Profiles
Create profiles and knowledge banks with memories, which get updated as you chat with CoreLLM. Similar to ChatGPT memories, but you actually own your data.
Frequently Asked Questions
How is CoreLLM different from Open WebUI?
CoreLLM is a Progressive Web App (PWA), so you can run it on your browser. You can also install it as an app on your device. Unlike Open WebUI, which requires Docker setup and technical configuration, CoreLLM is designed for ease of use and accessibility. You can use it offline, and all your data stays on your device without any complex configurations.
Who is CoreLLM for?
CoreLLM is for everyday users who value their privacy and want to keep their data secure. It's ideal for professionals, students, and privacy-conscious individuals who want to leverage AI without sacrificing control over their information.
How is CoreLLM different from ChatGPT?
CoreLLM keeps all your data on your device. Unlike ChatGPT, which sends your conversations to OpenAI's servers and retains them for weeks or months, CoreLLM processes everything locally. Your chats never leave your device, giving you complete control over your privacy.
What AI models will be available?
CoreLLM will support multiple open-source models including DeepSeek, Qwen, Gemma, Meta Llama, and more. Note that for larger models, you'll need to download Ollama to use them. You'll be able to switch between models based on your needs, all running locally on your device. In the future, we plan to add support for API key-based models like OpenAI's GPT-5 and Anthropic's Claude, allowing you to choose between local and cloud-based processing.
Can I use CoreLLM offline?
Yes! Once you install CoreLLM as a PWA, it works completely offline. Your conversations, files, and AI processing all happen locally, so you don't need an internet connection to use it.
When will CoreLLM be available?
We're working hard to launch CoreLLM soon. Join our waitlist to be notified as soon as it's ready. We'll also share development updates and early access opportunities with our waitlist members.
It's time to take back control of your data today.
Join the waitlist and be among the first to try.
No spam. Updates only when we have something meaningful to share. Pinky promise.