LM Mini Blog
Local AI, explained simply
Guides and stories on on-device AI, LM Studio, Ollama, Personas, and private chat — written by the maker of LM Mini.
Guides and stories on on-device AI, LM Studio, Ollama, Personas, and private chat — written by the maker of LM Mini.
Compare up to four local models on the same prompt with LM Mini Arena — or run a multi-model Group Chat room for debate and specialized roles. Works with LM Studio, Ollama, Jan, oMLX, and on-device.
Connect iPhone or Android to your home Ollama server over Wi-Fi (port 11434), then use VPN/tunnel or LM Mini for away-from-home chat — without exposing the API to the open internet.
LM Studio does not need internet to chat once a model is on your machine. Neither does Ollama. Downloads need the net. A phone away from home needs remote access, not a public LM Studio port.
For on-device AI in LM Mini, smaller quantized models (roughly 1B–8B, depending on RAM) usually feel best. Bigger is not always smarter on mobile — keep heavy models on LM Studio or Ollama.
Voice chat in LM Mini: speech-to-text in, text-to-speech out, continuous conversation and Call Mode. Pair Kokoro voices with Personas. Works on-device, with LM Studio, and with Ollama — still private.
Private AI means your chats stay on your devices, not a random company's servers. Why that still matters in 2026 — work notes, health questions, drafts — and how LM Mini stays local by default.
LM Mini Connect for Mac is deprecated. It keeps working through the end of 2026. The replacement is LM Mini Home — keep working on phone and Mac with realtime sync. Windows and Linux Connect stay current.
Group Chat in LM Mini is a multi-model room. Several Personas or models share one thread — turn modes, parallel replies, scenarios, and group voice. Local and private.
Personas in LM Mini are saved AI characters with name, look, voice, model, and instructions. Build a writing coach, tutor, or coding buddy — free, local, and private.
LM Studio remote access from a phone is two setups: Serve on Local Network on Wi‑Fi, or away from home with LM Mini Home / Connect. Ollama works the same. No port forwarding. No VPN.
On-device AI means the LLM runs on your phone, not in the cloud. Learn when to use on-device models vs LM Studio or Ollama on a home PC — and how LM Mini supports both.
LM Mini is a local AI and private AI chat app for iPhone and Android. Run LLMs on-device, or connect to LM Studio and Ollama — including remotely with LM Mini Connect.