- 🔥 Passion System — The AI adapts its writing depth as your conversation evolves. Six tiers from casual to transcendent. No other app does this.
- ✍️ Impersonate — AI writes your next message. Tokens stream live into the input field. Edit or send as-is.
- 💡 Smart Suggestions — Three context-aware actions after every response. Match scene intensity, never repeat.
- 🔐 100% Offline — Runs on Ollama. Nothing leaves your device. Ever.
- 🎭 Custom Characters — Build your own with passion speed control, or use the 5 built-in personas.
- 🌍 13 Languages — EN, DE, ES, ZH, FR, IT, PT, RU, JA, KO, AR, HI, TR
- 🎙️ Voice & Image — Piper TTS, Zonos voice synthesis, Stable Diffusion image generation.
# 1. Install Ollama → https://ollama.ai
# 2. Pull a model
ollama pull HammerAI/mn-mag-mell-r1:12b-q4_K_M
# 3. Launch AriaFirst launch walks you through everything — no terminal knowledge required.
Models by VRAM
| VRAM | Top Pick | Alternatives |
|---|---|---|
| 0-2 GB | gemma3:1b |
llama3.2:3b |
| 2-6 GB | HammerAI/smart-lemon-cookie:7b |
HammerAI/neuraldaredevil-abliterated |
| 6-12 GB ⭐ | HammerAI/mn-mag-mell-r1:12b |
vanilj/mistral-nemo-12b-celeste-v1.9 |
| 12-24 GB | HammerAI/cydonia-v3.1 |
huihui_ai/gemma3-abliterated:27b |
| 24 GB+ | HammerAI/l3.3-omega-directive-unslop-v2:70b |
fluffy/magnum-v4-72b |
Build from Source
git clone https://github.com/Gakuseei/Aria.git
cd Aria
npm install
npm run dev- Phase 1 — Passion System v3, chat engine overhaul, smart suggestions, full manual testing
- Phase 2 — Premium personas, chat improvements, profile pictures, chat search
- Phase 3 — Custom installer (Ollama included), tutorial refactoring, stats dashboard
- Phase 4 — UI overhaul, dark/light mode, new logo
- Phase 5 — v1.0: group chats, multimodal vision, landing page
- Phase 6 — v2.0: Full Rust rewrite with Dioxus. One binary, no Chromium.
Ko-fi supporters unlock Gold Mode — exclusive visuals and early access.
MIT License — LICENSE.md
Made with 🌹 by Gakuseei