ToolPiper

Instant local AI

Only for Mac

Free · In‑App Purchases

Mac

The Ultimate MacOS AI tooling — let Claude Code and other AI tools control your local inference, browser testing, & model management. Best-in-class MCP & CDP capability. ToolPiper is a local AI inference engine and model manager that lives in your Mac's menu bar. One click gets you from zero to running AI — no terminal, no Python, no API keys required. Open the built-in web interface, pick a workflow template, and start chatting, creating, or learning with models that run entirely on your hardware. SIX INFERENCE BACKENDS, ONE APP ToolPiper bundles six AI backends into a single native macOS app: - llama.cpp — Run open-source LLMs and vision models using quantized GGUF files on Metal GPU - Apple Intelligence — Access Apple's on-device language model with zero downloads on supported Macs - FluidAudio STT — Speech-to-text powered by Parakeet on Apple Neural Engine (25 languages) - FluidAudio TTS — PocketTTS text-to-speech on CoreML - MLX Audio TTS — Soprano, Orpheus, and Qwen3 TTS with voice cloning, accelerated on Metal GPU - CoreML Image Upscale — 4x photo upscaler trained on real photographs, runs on Neural Engine Every backend is managed automatically. No configuration files, no port conflicts, no dependency hell. VISUAL AI PIPELINE BUILDER Go beyond simple chat. ToolPiper serves ModelPiper.com's Angular app locally. Similar to no-code platforms, you can build real-time AI pipelines by connecting blocks: - Mic to Speech-to-Text to LLM to Text-to-Speech to Speaker (a fully local voice assistant) - Text to Multiple LLMs in parallel (compare model outputs side by side) - Chain any combination of text, speech, vision, and image blocks Data streams through the pipeline as it arrives — no waiting for one step to finish before the next begins. CURATED MODEL LIBRARY Browse and install models with one click. ToolPiper manages downloads from HuggingFace and stores everything in a central model directory: - Qwen3 0.6B — Tiny reasoning LLM, ideal for getting started - Llama 3.2 (1B and 3B) — Meta's compact, capable models - Qwen 2.5 (3B) — Multilingual LLM from Alibaba - DeepSeek R1 (8B) — Reasoning-focused model - Gemma 2 (2B) — Google's efficient small model - Qwen 2.5 VL (3B and 7B) — Vision-language models that understand images - Parakeet TDT V3 — Multilingual speech-to-text on Apple Neural Engine - Soprano TTS — Fast multilingual text-to-speech with 8 voices - Orpheus TTS — Expressive speech synthesis with emotional control - Qwen3 TTS — Multilingual TTS with voice cloning from a short audio sample - PurePhoto SPAN — 4x photo upscaler on Neural Engine COMPLETE PRIVACY Every model runs entirely on your Mac. No data leaves your hardware. No accounts, no telemetry, no cloud processing. Your conversations stay yours. LOCAL AI API ToolPiper exposes a standard chat completions API on localhost. Any tool that works with the industry-standard chat completions format can point to ToolPiper — use it with your own scripts, other apps, or development workflows. THE MODELPIPER ECOSYSTEM ToolPiper is the inference engine that powers a family of companion apps: - MediaPiper — Browser extension for Chrome, Firefox, and Safari. AI-powered image discovery, hover preview, and upscaling on any webpage. - VisionPiper — macOS menu bar app for screen capture with AI vision workflows. Record regions, export GIFs, and stream your screen directly into AI pipelines. - AudioPiper — macOS menu bar app for real-time audio mixing and recording. Capture mic, system audio, or individual app audio and route it through AI pipelines. All companion apps are free and connect to ToolPiper for on-device inference. REQUIREMENTS - macOS 26 Tahoe or later - Apple Silicon (M1 or later) required - 16 GB RAM recommended (8 GB minimum for smaller models) - Models range from ~500 MB to ~6 GB depending on size and quantization Terms of Use (EULA): https://www.apple.com/legal/internet-services/itunes/dev/stdeula/ Privacy Policy: https://modelpiper.com/privacy Term of Use: https://modelpiper.com/terms

  • This app hasn’t received enough ratings or reviews to display an overview.

Model Load Orchestration Watch models load and unload accurately. Pose / Mocap API Upscale Video Beta with PiperSR - An ANE (no GPU) 2x upscaler! MCP Server Integration ToolPiper now includes a built-in MCP (Model Context Protocol) server — connect AI coding assistants like Claude Code directly to your local models, browser automation, and testing tools. Enhanced Browser Automation New CDP-powered capabilities: network capture & interception, JS/CSS code coverage, performance tracing, storage management, autofill, and WebAuthn passkey testing. PiperTest Runner & Export Run visual tests server-side and export to Playwright or Cypress code. Unified Model Management Improved model lifecycle with memory profiling and resource-aware loading. Web Crawling Crawl web pages via Cloudflare Browser Rendering for RAG and data extraction.

The developer, R BENJAMIN RACICOT, indicated that the app’s privacy practices may include handling of data as described below. For more information, see the developer’s privacy policy .

  • Data Not Collected

    The developer does not collect any data from this app.

    Privacy practices may vary, for example, based on the features you use or your age. Learn More

    The developer has not yet indicated which accessibility features this app supports. Learn More

    • Seller
      • R BENJAMIN RACICOT
    • Size
      • 35.2 MB
    • Category
      • Developer Tools
    • Compatibility
      Requires macOS 26.0 or later.
      • Mac
        Requires macOS 26.0 or later.
    • Languages
      • English
    • Age Rating
      4+
    • In-App Purchases
      Yes
    • Copyright
      • © 2026 Ben Racicot