LLMster vs Local Inference OnDevice Agent

Local AI, in your pocket. LLMster is a fast, private iPhone and iPad client for models running on your LM Studio server. Stream replies, reasoning, tool activity, and image-based conversations from the models you host—in an app designed to feel at home on iOS. Private by design LLMster connects directly to the server you choose. There’s no LLMster account, telemetry, hosted chat service, or cloud relay in the middle. Your conversation library stays on your device, while model context stays with your configured server. External integrations run only when you choose to enable them. Your models, wherever you run them Keep your models on the machine built for them: a workstation, home server, or always-on computer. Run the LM Studio desktop app or LM Studio’s headless daemon, point LLMster at the host, and start chatting. No companion desktop client is required. Choose the right model without leaving the conversation: • Search or sort by name, size, or format, then pin favorites • See which models support vision, reasoning, or tool calls • Change models without starting a new chat Organized from the start Search chats by title, model, or folder. Browse recent conversations across every folder, then organize, rename, pin, or duplicate important threads. Move and protect your chats Share individual conversations as .llmster files and import them on another device while preserving messages, images, reasoning, tool history, and chat settings. Back up and restore your entire local chat library—including folders, pins, drafts, and attachments—with optional password protection. Backups do not include server addresses, authentication keys, or app-wide settings. LLMster also verifies transferred response context against your configured LM Studio server before continuing a chat. Vision, direct to your server Attach up to four screenshots, photos, or scanned pages to a message and send them to a compatible vision model. LLMster resizes and compresses images on your device before sending them, so the originals remain with you. Open any attachment and pinch to zoom when you need to inspect the details. Hands-free voice chat Talk naturally using on-device speech recognition and text-to-speech. Audio stays on your device; only the resulting text is sent to your configured server. Natural turn-taking and subtle haptics keep the conversation moving without touching the screen. Control each conversation Every chat can have its own: • Integrations already configured in LM Studio • Custom system prompt • Tool-call visibility • Live reasoning preference Stream reasoning as Markdown while the model works, or keep the interface quiet with a compact Thinking indicator and open the completed reasoning when you want it. Control response auto-scrolling, and choose System, Light, or Dark appearance with your preferred accent color. Rich responses, as they arrive LLMster renders streaming Markdown smoothly, including syntax-highlighted code with copy controls, inline and block math, tables, links, lists, and more. Export or share transcripts with their original Markdown intact. Keep going while you move on LLMster asks iOS to continue eligible responses after you leave the app. When the system starts continued processing, progress appears as a Live Activity on the Lock Screen and, on supported iPhones, in the Dynamic Island. When a background response completes, LLMster can notify you locally—without routing your conversation through a push server. Tap a notification or Live Activity to return to the correct conversation. Setup instructions are available at https://llmster.app/support

Your AI. Your phone. Nobody else's business. LocalInference runs powerful open AI models entirely on your device. Your conversations never leave your phone — no account, no cloud, no analytics, no tracking. Ever. PRIVATE BY ARCHITECTURE, NOT BY PROMISE • Everything runs on-device; works fully offline after a one-time model download • Zero data collection — we couldn't read your chats if we wanted to • API keys (optional) stored in the secure keychain, sent only to the provider you chose A REAL ASSISTANT, NOT JUST A CHATBOT • Tools: calculator, web search, date & time, webpage reading • Contacts & calendar integration — every write action asks you first • Ask about photos with vision models • Reads answers aloud with on-device voices BUILT FOR TINKERERS • 16 curated models from 0.5B to 8B — Qwen3, Llama, Gemma, DeepSeek R1, Phi, SmolVLM • Only models that fit your device's memory are offered • Live tok/s, time-to-first-token, and token counts per response • Thinking models show their reasoning in collapsible blocks • Bring your own keys: Anthropic, OpenAI, Google — or any OpenAI-compatible server (llama.cpp, LM Studio, Ollama) on your own network Built with Llama. Gemma models provided under the Gemma Terms of Use.