Local LLM Chat & Assistants

Polished native on-device LLM chat running Llama, Qwen, Gemma, Phi and more via an optimized mlc-llm engine with OmniQuant quantization. Fully offline, no subscription.

Polished cross-platform AI client with one-tap local MLX models (Llama, Qwen, Gemma), Apple Foundation Models, MCP/tool calls, system-wide translation, and BYOK cloud.

Offline LLM client that runs GGUF/MLX/CoreML/Apple Foundation models from Hugging Face, with on-device RAG over your PDFs/EPUBs, vision, and optional remote endpoints.

Feature-packed offline AI studio: local LLM chat (Gemma/Qwen/Llama), on-device image gen and editing, vision, neural TTS, and even a local AI API server.

Run open LLMs — Llama 3, Gemma 2, Phi 3, Qwen 2, Mistral — fully offline on Mac, iPad, and iPhone with private chats and customizable system prompts.

Private AI assistant running open models (Gemma, Phi-3, Mistral, Llama) fully on-device, plus on-device image inpainting and vision Q&A, optimized for Apple Silicon.

Private offline LLM chatbot with custom characters running GGUF/MLX/MNN models from Hugging Face on-device, plus Siri shortcuts and optional Ollama backends.

On-device assistant built on Apple's Foundation Models for chat, OCR, image description, live vision/audio and Image Playgrounds generation, all processed locally.

Allen Institute's fully open OLMoE mixture-of-experts model running entirely on-device with no cloud; a private chat and research playground for local LLMs.

Private on-device chat running Apple's Foundation Model or open models (DeepSeek R1, Llama 4, Gemma 3n, Phi-4) on the Neural Engine, with tools and voice mode.

Open-source, minimal LLM chat that runs Llama, Qwen, Gemma and more locally on Apple Silicon, or connects to any OpenAI-compatible endpoint like Ollama.

Clean Markdown chat client for self-hosted Ollama models; lets researchers quickly switch models and test prompts against their own local server.

Describe your mood in plain language and it surfaces matching movies and shows with a Match score and reason; developer says all AI runs on your iPhone.

Privacy-first offline chat running Gemma, Llama, DeepSeek, Qwen, or Mistral on-device, with MLX vision models and optional bridges to local-network LLM servers.

fullmoon runs open LLMs fully on-device, optimized for Apple silicon; free, open-source, cross-platform, with Shortcuts integration. ~114 ratings, 4.4 stars.

Runs open LLMs (DeepSeek, Llama, Qwen, Gemma, Phi, Mistral) fully on-device with Metal acceleration for private, offline chat across Apple platforms.

A full AI workbench running local offline models (Qwen3, Gemma 3, Llama-3.2, DeepSeek) plus cloud providers, with RAG over documents, memory, and custom agents.

Private offline AI chat running MLX models on-device, with local image generation and a RAG system for analyzing your PDFs and documents.

Dating chat coach that drafts replies in your voice, reads screenshots, ranks profile photos, and runs roleplay practice, all on-device with Apple Intelligence.

Russet runs downloadable models fully on-device with no accounts or tracking, plus an agent, offline text actions, PDF/image analysis, and TTS.

Free, open-source on-device chatbot running IBM Granite 3.1 (LLM/MoE) fully offline on Apple's Neural Engine—no internet, no accounts, no data leaving your device.

Privacy-first chat assistant running open models on-device via Apple's MLX and Ollama on the Neural Engine; fully offline after setup. ~37 ratings.

Account-free chat client that connects to fully local Ollama/LM Studio models or cloud, with voice in/out. Highly rated (4.9) though small sample.

Runs open GGUF LLMs (Qwen 3.5, Gemma 4) fully on-device with a RAM-aware catalog, local web search/browsing tools, and on-device vision input.

On-device LLM chat that downloads and switches between Llama, Qwen, and Gemma, with offline voice chat and private PDF/text document analysis.

AI explorer/encyclopedia of models and tools with a Foundation Model Chat that talks to Apple's on-device LLM for private conversations. ~16 ratings.

Private offline chatbot with a custom Metal 3 engine for Apple Silicon, running any GGUF model (Llama, DeepSeek, Gemma) plus vision and RAG, fully on-device.

iPhone AI assistant that offloads inference to your own Mac running LLM Pigeon Server, so messages stay in your private iCloud container and never hit an outside cloud.

Universal on-device chat app running Gemma 3, Phi 4 Mini, DeepSeek R1 Distill, Qwen 2.5, and Llama 3.2 fully offline, with iCloud-synced history.

MLX-based local chat running 15+ Apple Silicon models or any HuggingFace import, plus Apple Foundation Models, vision, and on-device image generation.

Private offline chatbot running open models (DeepSeek, Qwen, Llama, Gemma, Apple Foundation, GPT-OSS) on-device, with image analysis and inline image generation on Mac.

The mobile companion to Acorn: a self-contained offline personal AI that runs fully on-device with no cloud, accounts, or subscriptions, plus memory.

Local character-card chat interface running GGUF models via llama.cpp, fully offline with no accounts or servers — built for tinkerers.

Runs open LLMs (Gemma, Qwen, Llama, Phi) fully on-device via MLX Swift on Apple Silicon, with chat, vision, agent tools, and a model browser.

Private AI chat that runs a language model locally on the Neural Engine with no accounts, optionally augmented by anonymous live web search.

Turns your Mac into a private local AI server: run 500+ language/vision models, chat in a built-in web UI, and expose them over your network via MCP and local APIs.

On-device LLM client for Apple Silicon supporting both MLX and GGUF models across Qwen3, Gemma3, BitNet, ERNIE and EXAONE, including multimodal vision chat.

Private chat running open LLMs locally, with on-device Stable Diffusion image generation, a local RAG knowledge base, and optional web search.

Runs Qwen3, Gemma 3, DeepSeek-R1 and other GGUF models fully offline with vision, RAG, WhisperKit voice and optional web search.

Illustrated weather app whose Trip Forecasts add light on-device AI help via Apple Intelligence to compare cities and dates; AI role is modest.

Offline AI assistant running Apple Intelligence, bundled and downloadable .gguf models, or Ollama/Draw Things; deep Siri, Shortcuts and Watch integration. Cloud optional.

Private chat assistant running open models (LLaMA, Mistral, Phi, DeepSeek, Qwen, Gemma) 100% on Apple Silicon. One-time purchase, no subscriptions, no cloud.

Runs 18 open-source models (Gemma, Qwen3.5, Llama) on-device via Apple's MLX, plus doc chat, vision/OCR, and voice; works in airplane mode.

Turns your Mac into a private AI server via MLX, running Llama, Gemma, Qwen, and Phi locally and exposing an OpenAI-compatible API to your other devices.

Offline survival assistant running a full local LLM (Apple Intelligence, Phi-4 Mini, or Llama 3.3 8B) with RAG over Red Cross/FEMA/WHO knowledge.

Offline AI workspace that runs and switches between on-device models for writing, coding, creativity, and image analysis — no cloud, no tracking, no server wait.

Discover, download, and run open-source LLMs fully offline on iPhone, iPad, or Mac, with prompts and outputs staying on-device.

Native Mac front-end for Ollama with vision-model image analysis, MCP tool integration, and fine-grained sampling controls; everything runs locally, no cloud.

Offline AI companion that runs over 10 local models on-device for private, context-aware chat with no cloud dependency.

All-in-one AI workspace mixing on-device and cloud chat with rewrite tools, voice-first chat, AI to-dos and custom personas, offering Apple Speech or Whisper input.

One-click local LLM runner: download and chat with Llama, Gemma, Qwen, Phi and DeepSeek offline, with a chain-of-thought view and MCP support. No terminal needed.

Private on-device assistant running Llama, Gemma, Phi and Qwen on the Neural Engine, with vision models and Stable Diffusion image generation, offline.

Private Mac chat assistant running open-source LLMs on Apple Silicon via MLX, with PDF Q&A; works fully offline after a one-time model download.

Private Mac assistant running local LLMs on Apple Silicon with cited PDF/image analysis and optional live web synthesis; no cloud inference, no subscription.

On-device AI chat using Apple Intelligence or downloaded Hugging Face models (Qwen, Gemma), with vision models and calendar tools, fully offline.

Turns an Apple-silicon Mac into a private inference server for the LLM Pigeon iOS app, running built-in models or LM Studio/Ollama with CloudKit sync and optional web search.

Polished Mac chat workspace for running local LLMs already on your machine, with model management, file/image drag-in, voice input and context compaction.

Pools RAM across multiple iOS/macOS devices over Wi-Fi or USB to run local LLMs too big for one machine, with tensor and pipeline parallelism.

A clean chat client that auto-discovers Ollama/LM Studio servers on your local network, supports vision models and Markdown, keeping chats on-device.

One-click local LLM runner that loads open models (gpt-oss, Qwen3, Gemma3, DeepSeek, Llama) on Apple-silicon Macs for private, offline chat.

Native Mac AI suite with a chatbot plus 10 tools (summarize, grammar, translate, code review); history kept locally, no account.

On-device LLM chat running Llama, Qwen, DeepSeek, Gemma, Phi and Mistral via Apple MLX, with one-tap model downloads and benchmarks.

Russian dream interpreter pairing a large symbol dictionary and lunar calendar with AI that gauges a dream's mood; an on-device claim isn't explicitly stated.

Offline document-based AI workspace for chat, image generation, character chat, PDF Q&A and bulk text rewrite/translate; runs on-device with Apple Intelligence.

Private AI workspace that runs Qwen/Gemma locally with zero setup or connects to 300+ cloud models, with agentic tools and MCP.

Hollywood producer sim where an on-device machine-learning model trained on real box-office data renders every critic, audience, and box-office verdict.

Open-source, privacy-first local LLM chat that runs Ollama, MLX-optimized Swama, or built-in on-device models with token/sec analytics and a floating panel.

Space-exploration game that generates planets, missions, and maps offline using Apple's on-device Foundation Model; a novel use of local LLMs for procedural content.

Private AI companion with an expressive VRM avatar that bundles local LLM and TTS models for offline voice chat and document reading.

Native Mac/iPhone AI companion with continuous conversation; runs local Ollama or LM Studio, plus Apple Foundation Models offline.

Private ChatGPT alternative running 15+ open models (Qwen, Mistral, Gemma, Phi-4, Llama 3, DeepSeek) on-device, with vision and PDF chat.

AI chat sidebar that sees the webpage you're on; bring your own key for cloud models or connect fully local to Ollama, LM Studio, llama.cpp or vLLM.

Runs Llama, Gemma, Qwen, and DeepSeek fully offline on Apple Silicon via MLX for private chat, summarizing, and image analysis.

Private on-device AI assistant for chat, writing, and summarizing; runs a local model with optimizations like Metal acceleration and works offline after setup.

Multi-model AI chat aggregating 26+ cloud models that also runs Llama, Gemma, Qwen, and SmolLM locally for offline, private conversations.

Minimal LLM testing app for pros: chat with on-device Qwen3 fully locally, or reach Claude/ChatGPT/Grok via an anonymizing proxy; history stored locally.

A private companion chatbot built on Apple's on-device 3B Foundation model; works in Airplane Mode and never transmits chats off-device.

Apple's desktop OS, with Apple Intelligence delivering on-device generative features (Writing Tools, summaries, Notes transcription) on Apple silicon with privacy protections.

Private chat engine powered by Apple Intelligence for ideas, planning, and recipes; works offline and supports translation. Requires Apple Intelligence hardware.

Runs GGUF LLMs fully on-device with model download and switching, offline operation, and configurable tasks for writing, coding, and brainstorming.

Apple Silicon-native LLM chat and character roleplay that runs models fully offline with no accounts or subscriptions; pay once.

Private on-device chat assistant on Apple foundation models with Write/Summarize/Explain/Plan/Brainstorm modes, PDF/image import, and reusable outputs.

Private AI chat that runs entirely on-device with one-tap model installation and zero conversation data collection.

Private GPT that runs open models fully offline on-device, or connects to your OpenAI/Claude API keys when you want cloud power.

On-device chatbot that runs Llama, Gemma, Mistral, Qwen, and OLMo locally with no cloud, plus speech input; subscription-gated after a free trial.

A private chat assistant running a downloaded on-device model (~1-2 GB) that works fully offline after setup, with no account, cloud, or tracking.

A private chat assistant powered entirely by Apple Intelligence's on-device model: chat, proofread, summarize, and rewrite with no account or internet.

Off-grid mesh messenger with E2E-encrypted calls that also bundles a conversational assistant running 100% on-device via Apple Intelligence.

Private executive companion running a 7B local model ('Venus') fully offline with perpetual local memory; an optional privacy-first cloud tier adds heavier reasoning.

Offline macOS chat that runs Gemma models locally and imports custom .gguf files; fully local execution with no cloud uploads.

Private chat client powered by the efficient RWKV RNN architecture, running LLMs fully offline with model switching, custom system prompts, and local history.

Airbnb trip-planning companion with 22+ Apple Intelligence features (concierge, listing verdicts, itinerary doctor, receipt OCR) running on-device.

Account-free private AI chat for Mac that runs local models and works offline, with simple model tiers and support for Ollama, LM Studio, and llama.cpp.

Local-first macOS chat that runs core generation on-device via Apple's Foundation Models, with optional user-initiated web search and voice in/out.

Unified Mac client for GPT, Claude and Gemini that also runs local models via Ollama or Apple Intelligence on-device, with files, vision and transcription.

Private on-device chat that runs Apple Foundation Models with no download, or a curated catalog of local GGUF models (Qwen3, Phi-4, Gemma) you install yourself.

A compact, themeable chatbot UI for conversing with locally hosted models or Apple's Foundation Models, with customizable AI profiles and instructions.

Beginner-friendly toolbox that wraps local LLaMA-via-Ollama into four modes (Plan, Summarize, Decide, Write) with no terminal setup; everything stays on-device.

Unified BYOK chat front-end for GPT, Claude, Gemini plus local Ollama and Apple Intelligence models, with shared encrypted memory across intelligences.

Runs a quantized 3B on-device model fully offline with template-driven prompts, reusing the Apple Intelligence model so app size stays small; requires Apple Intelligence.

Unified AI workspace with agentic Thinking Mode and vision support, running local models via Ollama or connecting to Gemini and OpenAI cloud models.

Chatbot client for any LLM API or local model, with an on-device web-search pipeline and a fully local whisper-v3-large-turbo voice transcription mode.

A self-contained offline AI assistant for Mac that runs entirely on-device with no cloud, accounts, or subscriptions, and a personal memory feature.

Retro-themed chat engine powered by Apple Intelligence that works offline for story ideas, trip planning and recipes, with voice input.

A bring-your-own-keys Apple-native workspace for multi-agent "covens" mixing cloud providers with local inference via LM Studio, Ollama, and llama.cpp.

AI photography mentor that analyzes your photos across composition, exposure, and color, then guides reflection and builds a personalized practice plan.

Menu-bar workspace blending notes, kanban and a hybrid AI chat that routes to cloud flagships via OpenRouter or to private offline models through local Ollama.

Hybrid chat that runs Llama/Qwen/Gemma locally via llama.cpp offline, or connects to Gemini, Groq, and OpenRouter with your own keys.

On-device chatbot built on Apple's Foundation Models with structured templates for recipes, itineraries, and summaries; conversations never leave the device.

Lightweight privacy browser bundling an integrated 3-billion-parameter model the developer says runs entirely private on-device; rough around the edges.

Birding chat assistant that queries the eBird database via MCP and, on iOS 26+, can run an on-device model so conversations and summaries stay on your device.

Street-view geography guessing game where on-device AI gives progressive offline hints about your location without handing over the answer.

Chatbot that runs local LLMs and image-generation models with no API needed, keeping inference entirely on-device.

All-in-one native macOS AI workspace: chat, web summaries, and audiocasts, with optional on-device Apple Foundation Models alongside BYO-key cloud providers.

Customizable multi-provider AI chat client that also offers a curated set of local models to run on-device, with per-chat system prompts, files, TTS and image gen.

Native SwiftUI Mac chat app powered by Apple Intelligence, with streaming responses and keyboard-first navigation; every conversation stays on-device.

Menu-bar color assistant: describe a design context in plain English and Apple Intelligence generates a six-color palette with reasoning, fully on-device.

A minimalist word-guessing game where on-device AI picks a secret word and judges which of your two guesses is closer; no internet required.

Desktop AI companion that bundles a local LLM and runs a localhost server on launch; streaming chat with system-prompt and temperature controls, no API keys.

Friendly talking voice companion running a small on-device AI brain; after a one-time download it answers spoken questions offline, age-tailored.

A paid offline chatbot that runs a Qwen3.5-4B model entirely on-device, streaming tokens with no servers; needs 8GB+ RAM for smooth replies.

Bring-your-own-key multi-provider AI workspace (GPT, Claude, Gemini, more) that also runs local models offline, with PDF/image chat, OCR, and prompts.

Malaysia-focused transit assistant that uses NLP and a deterministic routing engine to answer mixed Malay-English questions across KTM, MRT, LRT, and ferry lines.

A private Mac chat workspace built on Apple Intelligence on-device models, with a Core Data timeline so every conversation is searchable and taggable.

Native Mac client that queries two or three models side by side with diff highlighting; supports local Ollama models alongside cloud APIs.

Native Mac micro-CRM for freelancers with timezone-aware client times and Apple Intelligence-generated business insights; all data stays local.

Multi-model LLM client built around Ollama so you can run open models locally and free; requires an installed Ollama server.

Fully offline AI assistant with voice input and multi-session management, running entirely without a network connection for privacy.

AI dictionary/encyclopedia powered by Apple Intelligence answering questions in English or Chinese with text-to-speech; works offline. Requires Apple Intelligence devices.

SoulTalk AI roleplay companion chatbot with custom personalities and emotional-tone matching; developer states chats stay private on your device.

Flight tracker whose Predictive Delay Score uses on-device intelligence to rate each flight from route, carrier, aircraft and weather, plus an AI-drafted compensation claim.

A setup helper that installs local LLMs on your Mac and launches OpenWebUI in your browser for fully private, ChatGPT-style chat with no data collection.

Multi-modal Mac AI assistant that switches between OpenAI, Anthropic, Gemini, and local Ollama models with a customizable canvas.

Lightweight client for chatting with locally hosted LLMs via API, with document-context querying and Markdown chat export. Optional OpenAI ChatGPT support.

Baby-name companion pairing a 2,000+ curated library with on-device Apple Intelligence that tailors personalized name suggestions privately.

Native Mac AI copilot you can summon anywhere to brainstorm, write, and learn; developer says it works locally and is more eco-friendly than cloud assistants.

Facebook Messenger client with Apple Intelligence AI Reply and live translation; AI runs on-device on macOS 26+ where supported.

Tiny LLM client that runs Apple Intelligence on-device or connects to LM Studio/OpenAI, with function-calling for calendar tasks. A free research demo.

Essential-oils reference and blend builder with dilution calculator, now adding an on-device AI suggestions beta for oil recommendations.

Pocket chat companion built on Apple Intelligence on-device models, saving conversations to a searchable local timeline with no servers or accounts.

Brand-naming studio with an on-device AI pronunciation/phonetic engine, conflict detection and scoring; optional external AI for logos and identity assets.

Kid-friendly learning chatbot with quizzes and story/image generation, using on-device Apple Intelligence when available and a cloud fallback on older devices.

Organic gardening companion with a 500+ plant knowledge base and an offline AI assistant for pest control, soil, and what-to-plant questions.

Timeshare sales toolkit pairing financial calculators with an Apple Intelligence chat that researches 60+ brands fully offline.

Field CRM for window-coverings trades with an AI-powered quote builder that handles complex made-to-measure jobs and works fully offline on-site.

Floating chat bubble that connects to a locally-run OpenClaw gateway for instant AI access across Mac spaces; communication stays local.

Lightweight app to chat with local AI models on device, transcribe audio in-app, and remotely reach your computer via SSH or VNC, with iCloud-synced history.

Ghost-hunting novelty spirit box that turns simulated device sensor input into AI-generated words spoken aloud via text-to-speech; entertainment only.

An AI Feng Shui room guide: scan a room, overlay a Bagua grid, get a Harmony Score plus layout, color, plant, and lighting suggestions.

Turns YouTube subscriptions into an auto-playing personal TV, using Apple Intelligence to generate culturally-aware multilingual queries for dubbed content.

Social party game of secret missions played across phones, with a Pack Forge that uses a built-in AI helper to generate custom mission packs. Mostly a game, AI is a minor feature.

Bare-bones app to run an LLM on your device via Apple Intelligence; minimal description and no ratings yet.

Lottery companion that generates number sets via Apple Intelligence or ChatGPT, checks draws, and stores picks; the 'AI' picks are essentially randomized.