🧠 Offline AI Chat & Coding GUI (Rakyat Edition)
🚀 Multitab Asli ≠ Multisession 🚀
Selama ini kamu pikir multitab = multisession? Salah besar.
Punya gue 1 otak AI jalan bareng di banyak tab, tanpa tabrakan, tanpa timeout.
Buka 3–5 tab? Tetep sinkron.
Model cuma 14B Q6_K_XL, RAM 16GB? Masih ngacir.
Queue request otomatis → siapa input duluan, itu yang jalan.
Bukan “tab bohongan” kayak di webUI lain. Ini multitab asli non-multisession.
Sekali nyobain, balik ke multisession bakal berasa jadul.
😏 Ada yang berani tes bedanya?
🎥 Video :
✅ FAQ — Pertanyaan Umum (Trust Booster Edition)
❓ GUI ini beneran bisa jalanin model 13B tanpa GPU? Ya! Sudah diuji langsung dengan model llama-2-13b-chat.Q4_K_M.gguf di sistem dengan:
💻 CPU: Intel i5-9400F (tanpa iGPU)
🧠 RAM: 16GB DDR4
⚙️ Backend: llama.cpp
📦 GUI: Llamacpp AI Chatbot GUI
❓ Bukti nyatanya mana? 📸 Screenshot saat load model dan idle sudah diunggah di folder docs/screenshots/
📄 Log lengkap sesi percobaan model 13B tersedia di docs/session-logs/
Tidak ada error, tidak crash, hanya delay wajar saat proses berat.
❓ GUI-nya berat gak? Tidak. GUI ini hanya 10KB, tanpa dependensi besar seperti Gradio atau Electron.
Tidak buka port aneh-aneh.
Tidak ada tracking.
Murni offline dan lokal.
UI sangat ringan, hanya berbasis tkinter.
❓ Bisa pakai model 7B, 8B, atau 13B lain? Bisa! Sudah diuji dengan:
Mistral 7B
DeepSeek Coder 6.7B
DeepSeek Coder 7B
Nous Hermes 13B (Q4_K_M)
LLaMA 13B (Q4_K_M)
❓ RAM saya cuma 8GB, bisa jalan? Bisa, asal model yang dipilih sesuai. Gunakan model kecil seperti:
TinyLlama 1.1B Q8_0
DeepSeek Coder 1.3b Q8_0
Mistral 4B Q8_0
Open Hermes 7B Q4_K_M
Atur max_tokens di GUI agar tidak melebihi kapasitas RAM kamu.
❓“Saya masih nggak percaya GUI ini bisa jalanin model 13B cuma dengan RAM 16GB. Beneran bisa?” 💬 “Coba sendiri aja bro 😎”
❓“Emang GUI-nya ringan banget ya?” ✅ Iya. Ukuran file .py cuma mb. Gak ada embel-embel web server, backend rumit, atau library berat.
❓“Bisa crash gak pas load model besar?” 🚫 Selama sistem kamu stabil dan swap file aktif, hampir nggak pernah crash. Bahkan log menunjukkan performa tetap normal walau RAM di atas 15GB pas awal load.
❓“Ada buktinya?” 📸 Sudah ada screenshot dan log di folder docs/session-logs/ dan docs/screenshots/.
❓“Kalau saya nggak percaya tetap?” 😎 Silakan buktikan sendiri.
✅ FAQ — Frequently Asked Questions (Trust Booster Edition)
❓ Can this GUI really run a 13B model without a GPU? ✅ Yes! Successfully tested with llama-2-13b-chat.Q4_K_M.gguf on:
💻 CPU: Intel i5-9400F (no iGPU)
🧠 RAM: 16GB DDR4
⚙️ Backend: llama.cpp
📦 GUI: Llamacpp AI Chatbot GUI
❓ Where’s the real proof? 📸 Screenshots during model load and idle are uploaded to docs/screenshots/ 📄 Complete 13B model session logs available in docs/session-logs/ ✅ No errors, no crashes. Just slight delay under heavy processing — perfectly normal.
❓ Is this GUI heavy? ❌ Not at all. It’s just 10KB. No bloated dependencies like Gradio or Electron. ✔️ No random ports. No tracking. ✔️ 100% offline and local. ✔️ Based purely on Tkinter.
❓ Can I use other 7B, 8B, or 13B models? ✅ Absolutely! Already tested with:
Mistral 7B
DeepSeek Coder 6.7B
DeepSeek Coder 7B
Nous Hermes 13B (Q4_K_M)
LLaMA 13B (Q4_K_M)
❓ I only have 8GB RAM, will it work? ✅ Yes, just use smaller models like:
TinyLlama 1.1B Q8_0
DeepSeek Coder 1.3b Q8_0
Mistral 4B Q8_0
Open Hermes 7B Q4_K_M
🛠️ Set max_tokens low to match your available RAM in the GUI settings.
❓ “I still don’t believe this GUI can run 13B on just 16GB RAM. Really?” 💬 “Try it yourself, bro. 😎”
❓ “Is the GUI really that lightweight?” ✅ Yep. File size is only mb. No web servers, no complex backends, no heavy libraries.
❓ “Will it crash when loading large models?” 🚫 As long as your system is stable and swap file is active, crashes are extremely rare. 📊 Even with RAM above 15GB during model load, logs show stable performance.
❓ “Is there actual proof?” 📸 Yes. Screenshots and logs are available in the docs/session-logs/ and docs/screenshots/ folders.
❓ “What if I still don’t believe?” 😎 Feel free to test it yourself.
