Home

📚 Termux AI Assistant Wiki (v2.0)

🌟 Core Features

• 10 Free AI Models via OpenRouter API
• Interactive Terminal UI with color support
• Encrypted API Key Storage in ~/.config/termux_assistant/
• Conversation History with timestamp
• No Root Required

🚀 Installation

curl -sL https://is.gd/Termux_Ai | bash

Auto-installs dependencies:

• curl - API requests
• jq - JSON processing
• bc - Math calculations

---

🧠 Model Catalog

🔍 Available Models

Model ID	Provider	Parameters	Best For
deepseek-r1	Deepseek	7B	General Q&A
qwen2.5-vl-32b	Alibaba	32B	Multimodal
rogue-rose	Sophosympatheia	103B	Technical analysis
Full list...

💡 Tip: Use mistral-small for fastest responses

---

🖥️ Usage Guide

🏁 First Run

1. Launch with termux-ai
2. Navigate to:
   Settings (2) → API Key (1)
3. Enter your OpenRouter key

🗣️ Chat Session

[MAIN MENU]
1. Start Chat Session  
2. Settings  
3. View History  
4. Exit

• Type exit or quit to end session
• Press Ctrl+C for emergency exit

⚙️ Settings

Option	Description	Default
API Key	OpenRouter authentication	-
Model Selection	Switch between 10 models	deepseek-r1
Max Tokens	Response length limit	800
Temperature	Creativity control (0.1-2.0)	0.7

---

🔧 Troubleshooting

🚨 Common Issues

Symptom	Solution
API errors	Verify key at OpenRouter
Model timeout	Switch to lighter model (gemma-3-27b)
Blank responses	Check debug.log for errors

📜 Log Files

# View error logs
cat ~/.config/termux_assistant/debug.log

# Check history
nl -ba ~/.config/termux_assistant/history

---

❓ FAQ

❔ How to update?

pkg upgrade termux-ai  # If installed via repo

or

curl -sL https://is.gd/Termux_Ai | bash  # Fresh install

❔ Where is data stored?

~/.config/termux_assistant/

• config - Settings
• history - Chat logs
• api_key - Encrypted credential

❔ Can I use paid models?

Yes - add funds to your OpenRouter account for GPT-4/Claude access

---

⚠️ Limitations

• No pipe input (echo "hi" | termux-ai)
• No command arguments (all settings in-menu)
• No background mode (keep Termux open)

---

📌 Pro Tips

1. Use termux-wake-lock for long sessions
2. Clear history via Settings → Clear History
3. Lower max_tokens on slow connections

---

🔗 Official Repository: Anon4You/Termux-Ai
📅 Last Updated: July 2024
🐞 Report Issues: GitHub Issues

---

Appendix: Detailed Model Specs

Model ID	Context Window	Supported Languages	Rate Limits
deepseek-r1	4K	EN/CN/JP	5 req/min
qwen2.5-vl-32b	8K	Multilingual	3 req/min
rogue-rose	16K	EN/DE/FR	1 req/min

---