MLX optimization + Tailscale encrypted mesh networking: focusing on both privacy and computing power. Locally, this wave completes the Apple Silicon local AI ecosystem.

LM Studio officially launches the mobile version Locally, supporting iPhone/iPad, integrating local operation with LM LINK remote computing power access. Locally is optimized for Apple Silicon, based on the MLX framework, supporting open-source models such as Llama 3.2, Gemma 4 series, Qwen 3, DeepSeek, and others, with the mobile device able to directly run Gemma 4 (E2B) and import GGUF models. Fully offline and privacy protected. To enable LM LINK, you need to authorize with the same account on both your phone and computer, establish an encrypted network via Tailscale, and remotely connect to your home/office desktop to run large parameter models.