Local AI Apps and Tools for Mac

Harbor helps people experiment with local LLMs by connecting model backends, chat interfaces and supporting services. Its CLI and companion app manage those connections in a shared development environment.

Best Claude HUD keeps session and repository information visible for developers working in Claude Code's terminal interface. It helps you check the active model, available context and working tree without leaving the conversation.

LocalVocal turns speech into live captions inside OBS for streamers and people recording video. It processes speech on your computer, so transcription can work offline without sending audio to a cloud service.

Pico MLX Server is a macOS app for people who want to run MLX models locally and use them with an existing AI chat client. It gives MLX Server a graphical interface and exposes local models through an OpenAI-compatible API. The project is open source under the MIT license and requires macOS Sonoma or later.

Contrabass coordinates AI coding agents working through project issues, for developers managing several tasks or agent teams. It runs locally on macOS and Linux and reimplements OpenAI's Symphony approach.

Albert coordinates Claude Code agents for developers who want to hand off a coding goal that spans several tasks. It plans the work and continues through successive rounds until it reaches the goal or a requested review checkpoint.

ZettaBrain RAG is a self-hosted document assistant for individuals and small teams working with policies, contracts and internal manuals. It helps them find answers and draft documents from files they already keep on their own infrastructure.

Real-CUGAN NCNN Vulkan is a command-line AI image upscaler for people enlarging anime artwork on their own computer. It uses Real-CUGAN models through the ncnn runtime to increase image resolution and reduce noise.

Nanocoder is a terminal AI coding assistant for developers who want to choose which model works on their code. It can use local models through Ollama or send requests to a hosted provider through an OpenAI-compatible API.

Agency Orchestrator coordinates specialist AI roles for developers and solo builders working on product plans, research or content. A task description can become a team that divides the work and combines its results.

Ultimate Vocal Remover separates vocals and other parts of songs on your computer. It's for people preparing karaoke tracks or isolating audio for music projects.

Claude Code Agent Monitor gives developers one place to follow coding agents and inspect their work across Claude Code, Cursor and Codex. It keeps conversations and activity available for review after a session ends.

FastChat is a self-hosted LLM platform for developers and researchers who want to serve chat models on their own hardware and compare their responses. It powers Chatbot Arena and includes a locally hosted side-by-side comparison interface. The Python code is open source under Apache 2.0; model licenses apply separately, including Llama's license for Vicuna.

sonible smart:comp is a proprietary audio compressor plug-in for producers, mix engineers and sound designers working on macOS or Windows. It analyzes your audio and maps suitable compression settings in a visual Compression Matrix, so you can compare compression characters while keeping access to conventional controls.

SwarmUI is a self-hosted AI image generation interface for people who want both straightforward generation controls and access to detailed ComfyUI workflows. You can use it with models running on your own hardware.

total-agent-memory gives AI coding agents persistent memory of decisions, fixes and project conventions across sessions and repositories. It's for developers who don't want to keep explaining earlier work to Claude Code, Codex CLI, Cursor or another MCP client.

FunASR is a speech-to-text toolkit for developers building transcription services and voice applications. It supports offline recordings and live speech on local hardware or self-hosted servers, with capabilities that depend on the model and runtime.

BrowserPilot is a Python browser automation agent for developers who want to describe web tasks in plain English. It controls Chrome through Selenium on your machine, or a remote browser through Selenium Grid. The software is open source under the MIT license, but its AI capabilities use OpenAI's cloud API and require an API key and internet access.

Codewhale is a coding assistant for developers who want an agent to inspect projects, make changes and check the results. It runs on your computer and lets you choose the model and the permissions it works under.

Codanna helps developers and AI coding agents explore a repository and assess which code a change could affect. It connects code search with function relationships, so an assistant can get context around a match in one response.

sonible prime:vocal is a proprietary desktop AI audio processor for singers, producers and editors who need to repair vocal or speech recordings before mixing. Its processing is trained specifically for vocals, with tools for cleaning up phone recordings, home demos and dialogue captured in untreated rooms.

Enchanted is a ChatGPT alternative for people who want an Apple app for chatting with models they host themselves. It requires your own Ollama server, which runs the models on your computer or a server you control.

DevSpace is a self-hosted MCP server for developers who want ChatGPT, Claude or another compatible client to work in projects on their own machine. It connects an assistant to selected local folders and the project's terminal tools.

Atomic Agent is a local AI agent for people who want an assistant to work with their browser, files and code on their own hardware. Its model and task loop can run offline, while sessions and memory stay on disk. It's free and open source under the MIT license.

ShellGPT is an AI terminal assistant for developers and others who want help writing commands and code without leaving the shell.

LabelMe is a desktop image annotation app for people preparing computer vision datasets. It combines manual drawing with model assistance for labeling objects and outlining image regions.

Open Interpreter is a locally installed terminal coding agent for developers on macOS, Linux and Windows. It's a fork of OpenAI's Codex that focuses on adapting agent behavior to lower-cost models. Commands run on your computer inside native sandboxing; model requests go to your selected provider, so local installation alone doesn't mean local inference.

Aider is a terminal coding assistant for developers who want an LLM to make changes in a new or existing project. It works with local models and cloud services, and keeps code edits reviewable through Git.

Surf is an AI notebook for people researching across webpages, videos and documents. It brings source material and writing into one workspace so you can ask questions about what you've collected while taking notes.

Chat2DB Community is a free local database client for developers, database administrators and analysts who want AI assistance beside their SQL editor. It runs on Windows, macOS and Linux, or in Docker with browser access. You connect your own AI model; where AI requests go depends on that connection, so the local app alone doesn't guarantee local inference.

pgvectorscale extends pgvector for developers who want to search large embedding datasets within PostgreSQL. It adds indexing and compression for AI applications that keep vector data in their existing database.

lm-evaluation-harness is a Python framework for researchers and developers comparing language models on shared benchmarks. Public prompts help make results comparable across model checkpoints and research papers.

Spark-TTS is a local text-to-speech system for developers and researchers building speech applications. It can reproduce a voice from reference audio or create a synthetic speaker with chosen vocal traits.

Remembra Relay saves coding session handoffs so developers can switch assistants or resume a project on another machine without reconstructing its progress. The next agent receives a brief of completed work, unfinished tasks and failing checks.