Hallucinatron: LLM Chat Client

Hallucinatron is an Android chat client for anyone running their own
AI models — or just tired of apps that phone home more than the AI does.

Connect it to Ollama, LM Studio, llama.cpp, or any OpenAI-compatible endpoint
and start chatting. Local network, private server, or cloud API — your call.

BRING YOUR OWN MODEL
Add any number of backends, each with its own base URL, API key, system prompt and other settings. Models are fetched live from the server.

Works with Ollama, LM Studio, llama.cpp, OpenAI, Mistral, Groq, DeepSeek, and
anything else that speaks the OpenAI API format.

BUILT FOR REAL CONVERSATIONS
- Streaming responses with a stop button
- Edit any message and regenerate from that point
- Per-conversation system prompt overrides
- Collapsible thinking/reasoning display (DeepSeek-R1 and tags)
- Quote-reply by long-pressing any message
- Export chats as Markdown

STAY ORGANIZED
- Pin, tag, and search conversations
- Prompt templates for reusable snippets
- Automatic AI-generated conversation titles (accuracy may vary)

WORKS IN THE BACKGROUND
Waiting on a slow local model? Lock your screen — you'll get a notification
when the response is ready, with a tap straight back to the conversation.

PRIVACY
Hallucinatron contains no analytics, no telemetry, and no ads. It
makes no network requests except to the AI endpoints you configure yourself.
We have no idea who you are, and we'd like to keep it that way.