LMSA for LM Studio & Ollama

TechMitten LLC

Contains adsIn-app purchases

About this app

LMSA – The Privacy First AI Chat App for LM Studio, Ollama & OpenRouter on Android

Turn your Android phone or tablet into a powerful AI assistant. LMSA connects to local AI servers like LM Studio and Ollama, or cloud models via OpenRouter, so you can chat with the world's best large language models — privately, securely, and completely on your terms.

If you're running DeepSeek, Llama 3, Mistral, Phi, Qwen, Claude, or Gemini, LMSA is the fastest, cleanest, and most private way to access AI on Android.

🔒 PRIVATE BY DESIGN — YOUR AI, YOU ARE IN CONTROL
• On-Device Encryption: Your local chat database is fully encrypted on-device, ensuring absolute privacy for your conversations.
• Privacy First: We will never track your messages. All messages are stored locally on your device.
• Local-First: All conversations stay on your device and local network.
• Direct Cloud Access: OpenRouter requests go straight to the provider — no middleman servers.
• Biometric Lock: Protect your chats with fingerprint or face unlock.

🆓 FREE FEATURES INCLUDE:
• Unlimited Chat: No daily limits for your core conversations.
• Local & OpenRouter Backends: Connect easily to Ollama, LM Studio, or OpenRouter.
• Custom Endpoint Usage: Connect directly to any compatible AI server.
• Core Web Search: Stay informed with 2 live web searches per day.
• File Attachments & Text-to-Speech: Feed documents locally and listen to responses.
• Auto Discovery: Scan your local network to automatically locate your local server configuration.

👑 UPGRADE TO LMSA PREMIUM (ONE-TIME LIFETIME LICENSE)
Unlock the complete LMSA experience with zero distractions, no subscriptions, and full feature access forever:
• 🎨 Unlimited AI Image Generation: Bring your imagination to life using the simple /image prefix directly in chat.
• 🌐 Unlimited Brave Web Search: Fetch unlimited real-time data, current news, and up-to-date information.
• 🎛️ Full Parameter & Context Control: Customize assistant behavior by editing system prompts and adjusting max tokens to control response length.
• 🚀 Save Custom Endpoints: Save and quickly switch between multiple server profiles.
• 👥 Templates & Personas: Create and chat with custom AI characters.
• 📥 Import/Export Saved Chats: Easily back up and move your local conversation history.
• 🔌 True Offline Mode: Run completely offline with local models on your own trusted network.
• 🗣️ Premium TTS: Unlock all expressive, natural voices.
• 🚫 Ad-Free Experience & Premium Support: Zero distractions and priority email assistance.

🤖 COMPATIBLE WITH LEADING LLMS
• Full support for GGUF models on LM Studio (PC/Mac).
• Full support for Ollama servers and compatible models.
• Instant access to 100+ cloud LLMs via OpenRouter with your API key.
• Thinking Mode: See the full reasoning chain for deep reasoning models like DeepSeek-R1.

🛠️ EASY SETUP IN MINUTES
For Local AI (LM Studio / Ollama):

1. Install LM Studio or Ollama on your PC or Mac.

2. Load a model and start the local server.

3. Connect your phone to the same Wi-Fi network.

4. Enter the server IP and port in LMSA — start chatting.

For Cloud AI (OpenRouter):

1. Get an OpenRouter API key at openrouter.ai.

2. Enter your key in LMSA settings and start chatting instantly.

🧑‍💻 PERFECT FOR:
• AI enthusiasts experimenting with local language models
• Privacy-focused users who refuse to send data to big tech
• Developers, researchers, and power users building custom workflows

📋 REQUIREMENTS:
• Android 6.0 or higher
• LM Studio/Ollama on PC/Mac (on the same Wi-Fi) OR an OpenRouter API key

⚠️ Notice: LMSA is an independent app, not affiliated with LM Studio, Ollama, or OpenRouter. Local connections may be unencrypted — use only on trusted networks.

Private by design. Powered by you. Experience AI without the middlemen—download LMSA today.

Ratings and reviews

Pretty good overall. Better than most. Good tools integration.A few things: - Whatever it is sending for thinking mode is driving qwen (3/3.5) insane. Goes on and.on with "wait, I should" for dozens of sentences before finally selecting a reply. Something I don't see from other clients - UI has a few odd quirks still, like wanting to be completely full screen - For a premium app I would like to be able to control morez like controlling the thinking budget, see more about what is being sent.

The application works. You can't change the context length. It's trapped at 8,000 using local. No setting to increase context makes it unusable. Looked all over the app for it, couldn't find it. Also, models seem to be loading twice once when selecting the model of choice the other when chatting. When chatting locally, it seems to lock at 8,000 in context length. Adjusting it on the PC doesn't work. As soon as you start talking, it immediately resets back to 8000.

good app, does what it says on the tin. the UX needs some TLC (creating custom templates, preventing the new line button from sending messages, making the popup confirming the model is loaded show up less often/ less intrusively) but the foundation is rock solid. The cost to remove ads is extremely cheap, making this a solid all- rounder. looking forward to future updates!

What’s new

We squashed a few bugs and made some improvements to the chat input field in this release.