Locally AI
Locally AI is an on-device AI application that allows users to run powerful language models directly on their iPhone, iPad, or Mac without relying on cloud infrastructure or an internet connection. Built on Apple’s MLX framework, it delivers fast, efficient performance while minimizing power usage, enabling a seamless experience for chatting, creating, learning, and exploring AI capabilities across devices. It supports multiple open models such as Llama, Gemma, Qwen, and DeepSeek, allowing users to switch between them and tailor outputs to different tasks. Everything runs entirely offline, meaning no login is required, and no data is collected or transmitted, ensuring complete privacy and control over personal information. Users can interact with AI through natural conversations, analyze documents or images, and generate text in a unified interface designed for simplicity and responsiveness.
Learn more
WebLLM
WebLLM is a high-performance, in-browser language model inference engine that leverages WebGPU for hardware acceleration, enabling powerful LLM operations directly within web browsers without server-side processing. It offers full OpenAI API compatibility, allowing seamless integration with functionalities such as JSON mode, function-calling, and streaming. WebLLM natively supports a range of models, including Llama, Phi, Gemma, RedPajama, Mistral, and Qwen, making it versatile for various AI tasks. Users can easily integrate and deploy custom models in MLC format, adapting WebLLM to specific needs and scenarios. The platform facilitates plug-and-play integration through package managers like NPM and Yarn, or directly via CDN, complemented by comprehensive examples and a modular design for connecting with UI components. It supports streaming chat completions for real-time output generation, enhancing interactive applications like chatbots and virtual assistants.
Learn more
oMLX
oMLX is a macOS-native MLX server designed to make local AI faster and more practical on Apple Silicon. Built for the way coding agents actually work, it uses paged SSD KV caching to persist cache blocks to disk, allowing previously seen prefixes to be restored across requests and server restarts instead of being recomputed from scratch. This can reduce time to first token on long contexts from 30–90 seconds to under five seconds after the first turn. Continuous batching handles concurrent requests through mlx-lm’s BatchGenerator, improving generation throughput without forcing requests to wait behind a single job. oMLX can serve LLMs, vision-language models, embedding models, and rerankers simultaneously, using LRU eviction when memory runs low. It supports any MLX-format model from Hugging Face, including Qwen, LLaMA, Mistral, Gemma, DeepSeek, MiniMax, and GLM, and can reuse models already stored in the standard Hugging Face cache, LM Studio folders, or custom directories.
Learn more
Macyou
Macyou rents dedicated Apple Silicon Macs for AI workloads. Users configure a Mac (M4 Mac mini to Mac Studio M3 Ultra with 256 GB unified memory), pick a pre-configured stack — local LLMs via Ollama (Llama, Qwen, Mistral, DeepSeek), agent frameworks (CrewAI, LangGraph), or ML dev environments (MLX, Jupyter) — and get a running deployment in about 5 minutes. Every deployment exposes an OpenAI-compatible API, so existing OpenAI SDK code works by changing base_url; access also includes SSH with root and a browser-based remote desktop. Each customer gets a dedicated physical machine with full-disk encryption and a disk wipe between tenants, hosted in a GDPR-friendly jurisdiction. Pricing is a fixed monthly fee per machine with no per-token charges; Thunderbolt 5 clustering pools unified memory across nodes for larger models. Published, measured inference benchmarks (raw JSON, CC BY 4.0) show real tokens per second per chip.
Learn more
