Skip to content
Back to search

vesta-mac-dist

Vesta macOS Distribution - Official releases and downloads.Vesta AI Chat Assistant for macOS - Built with SwiftUI, Swift MLX and Apple Intelligence using Apple's on device model on MacOs Tahoe (MacOS 26). Now with side-by-side Qwen3-VL for vison

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/scouzi1966--vesta-mac-dist/badge.svg)](/listings/scouzi1966--vesta-mac-dist)

HTML

<a href="/listings/scouzi1966--vesta-mac-dist"><img src="/api/listings/scouzi1966--vesta-mac-dist/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

brew tap scouzi1966/afm
brew install --cask scouzi1966/afm/vesta-mac

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 245 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # Vesta AI Explorer for macOS - macOS Tahoe 26
  2. ### [vesta-mac.pages.dev](https://vesta-mac.pages.dev)
  3. ## Install with Homebrew (recommended)
  4. ```bash
  5. brew tap scouzi1966/afm
  6. brew install --cask scouzi1966/afm/vesta-mac
  7. ```
  8. ## Direct Download
  9. | Channel | Version | Download | Date |
  10. |---------|---------|----------|------|
  11. | **Stable** | 0.9.7 | [Vesta-0.9.7.dmg](https://github.com/scouzi1966/vesta-mac-dist/releases/download/v0.9.7/Vesta-0.9.7.dmg) | 2026-05-08 |
  12. | Nightly | Vesta-next | [Vesta-next.dmg](https://github.com/scouzi1966/vesta-mac-dist/releases/download/Vesta-next/Vesta-next.dmg) | Vesta-next |
  13. ### Nightly via Homebrew
  14. ```bash
  15. brew install --cask scouzi1966/afm/vesta-mac-next
  16. ```
  17. Vesta is a multi-backend AI chat application for macOS that runs models locally on Apple Silicon. It supports five AI backends simultaneously -- Apple Intelligence, MLX, llama.cpp, HuggingFace Inference API, and OpenAI-compatible API servers -- with runtime switching, vision understanding, text-to-speech, speech-to-text, image generation, video generation, and a full MCP server for programmatic control.
  18. ### Demo
  19. [![Vesta Demo](demo-preview.gif)](https://vesta-mac.pages.dev/mcp-demo.mp4)
  20. *Click to watch the full demo video*
  21. Note that there is a known bug in Qwen3-VL models where it will loop indefinitely in some cases and never complete. In this case stop the generation and tweak your prompt. This is a model behavior, not the app.
  22. Reference: https://github.com/QwenLM/Qwen3-VL/issues/1611#issuecomment-3639175711
  23. ---
  24. ## Vesta 0.9.7 (May 8, 2026)
  25. MLX model support sync with maclocal-api and release pipeline hardening.
  26. ### Changes
  27. - **Qwen 3.6 support** -- imported maclocal-api architecture mapping for Qwen 3.6 MLX models.
  28. - **Gemma 4 support** -- aligned Gemma 4 model handling and chat template fallback behavior with maclocal-api.
  29. - **Crash fix** -- corrected Gemma 4 VL attention reshape handling for prompt generation.
  30. - **Release publishing** -- rebuilt from scratch in Release, notarized, stapled, and published as a signed DMG.
  31. ---
  32. ## Vesta 0.9.6 (Mar 4, 2026)
  33. Stability and release pipeline improvements.
  34. ### Changes
  35. - **Stacked download button** -- version and date displayed vertically on the website download button
  36. - **Publish date on stable button** -- stable download button now shows the release date
  37. - **Expanded test suite** -- automated test pipeline expanded to 4 stages with full coverage
  38. - **Pre-test validation** -- release skill checks for running Vesta instance before tests
  39. ---
  40. ## Vesta 0.9.5 (Feb 11, 2026)
  41. **94 commits, 2 months of development since v0.9.0**
  42. This is a major release that transforms Vesta from a two-backend chat app into a full multi-modal AI platform with five backends, MCP integration, and media generation capabilities. Explore the world of AI beyond ChatGPT!
  43. ### New in 0.9.5
  44. #### Multi-Backend Architecture
  45. - **HuggingFace Explorer** -- Leverage the HuggingFace Pro subscription and inference providers partners for ImageGen, Videogen, Image edit, Whisper etc. I am not affiliated with HuggingFace but the pro sub is worth it.
  46. - **5 simultaneous backends** -- Apple Intelligence, MLX, llama.cpp, HuggingFace Explorer, and External AI (OpenAI-compatible) all active at once (Single chat routing)
  47. - **Runtime backend switching** -- switch between backends without restarting the app
  48. - **Per-backend settings** -- each backend has its own generation parameters, model selection, and configuration
  49. - **HuggingFace Browser** -- Direct browse and download models from HuggingFace in-app
  50. #### MCP Server (Model Context Protocol) - Agentic Sidekick!
  51. - **Full MCP server** running on TCP loopback with token-based authentication
  52. - **Agentic Sidekick** Vesta detects Claude Code when you enable MCP - Claude Code acts as an agent with a NLI (Natural Language Interface) to the app. Ask Claude to set things up, have a conversation with any other model!
  53. - **41+ tools** -- backend management, chat, model download/load/unload, vision analysis, settings, conversation history search, diagnostics, UI navigation
  54. - **6 resources** -- app state, models, conversation, settings, logs, system info
  55. - **7 prompts** -- guides for Vesta, MLX, llama.cpp, HuggingFace, and common workflows
  56. - **AI Sidekick** -- Claude Code integration for programmatic Vesta control
  57. - **Conversation history search** -- full-text search and read-only SQL queries against the SQLite message database
  58. #### HuggingFace Explorer (New Backend)
  59. - **Cloud inference** via 16+ providers (Cerebras, Groq, Together, Fireworks, SambaNova, Nebius, Replicate, and more)
  60. - **Text-to-image generation** -- FLUX.1 Schnell/Dev, FLUX.2, Stable Diffusion 3.x, SDXL with configurable size, guidance, and steps
  61. - **Image editing** -- instruction-based editing with FLUX Kontext and FLUX.2 via Replicate
  62. - **Video generation** -- Wan 2.2 T2V (text-to-video) with async polling
  63. - **Speech-to-text transcription** -- OpenAI Whisper models via HuggingFace Inference API with 14 languages
  64. - **Vision/VLM** -- Qwen2.5-VL, Qwen3-VL and other vision-language models via cloud
  65. - **Model browser** -- search and discover models from HuggingFace Hub with download counts, likes, and gated model detection
  66. - **Reasoning display** -- chain-of-thought rendering for models that emit `<think>` tags (DeepSeek R1, QwQ, etc.)
  67. - **7-tab settings panel** -- Chat, Vision, Image, Edit, Transcribe, Video, Settings
  68. #### Text-to-Speech (TTS)
  69. - **Kokoro** (82M) -- 46+ voices across 10+ languages, fast and high-quality
  70. - **Marvis** (100M/250M) -- conversational TTS with voice cloning support via reference audio
  71. - Models download on first use from HuggingFace
  72. #### Speech-to-Text (STT) -- WhisperKit
  73. - **On-device transcription** via WhisperKit CoreML -- zero network required
  74. - **6 model sizes** -- Tiny (39M) through Large V3 (1.5B) and Large V3 Turbo (809M)
  75. - **28+ languages** with auto-detect
  76. - Per-segment timing and speed ratio reporting
  77. #### Jinja Template Support (minja)
  78. - Full Jinja2 template parsing for GGUF models via llama.cpp's minja library
  79. - Correctly renders chat templates embedded in model metadata
  80. - Falls back to llama_chat_apply_template() for non-Jinja templates
  81. #### GGUF Model Browser
  82. - Search and browse GGUF models from HuggingFace Hub
  83. - Capability badges: Vision, Tool Use, Reasoning, Coding, Math, Multilingual
  84. - Split/multipart file detection
  85. - Automatic mmproj detection for vision models
  86. #### MLX Improvements
  87. - **Qwen3-VL M-RoPE patch** -- +81% performance improvement for vision inference (auto-applied via script)
  88. - **KV cache controls** exposed -- max KV size, quantization bits, prefill step size
  89. - **MLX benchmark tool** (mlx-bench) for standalone performance testing
  90. - **Wired memory** set to 90% of GPU recommended working set for large model performance
  91. ### Improvements
  92. - **Per-message metrics** -- token count and tokens/sec stored with each message in the database
  93. - **Green parameter labels** -- visual indicator when a generation parameter matches the model's configured default
  94. - **Binary voice format** -- Kokoro voice files converted from JSON to binary (144 MB down to 27 MB)
  95. - **Conversation history view** with backend filtering and pagination
  96. - **One-command build** -- `build-from-scratch.sh` handles submodules, patching, llama.cpp library build, and Xcode build
  97. - **Distribution pipeline** -- automated DMG creation, notarization, and GitHub release via `build-vesta-mac-dist.sh`
  98. - **Automated testing framework** -- 44+ UI tests via MCP-based test runner
  99. ### Bug Fixes
  100. - Fix AVKit VideoPlayer crash during SwiftUI transitions (disabled transition animations)
  101. - Fix download progress stuck at 0% and crash in llama.cpp streaming
  102. - Fix TTS mode hijacking text generation when both TTS and LLM models are loaded
  103. - Fix O(N^2) reasoning parser performance
  104. - Fix reasoning parser stripMarkers bug and chat history contamination
  105. - Fix streaming throttle not kicking in when content scrolls off-screen
  106. - Fix llama.cpp default context size (2048 changed to 16384 for Qwen3-VL)
  107. - Fix Continuity Camera Swift 6 concurrency crash in Release builds (Objective-C workaround)
  108. - Fix NSHostingView constraint crash in MLX settings window (non-observing wrapper)
  109. - Fix MoE warmup crash for models with 32+ experts (reduced warmup batch size)
  110. - Fix mxfp4 MoE Metal shader crash (skip warmup for mxfp4 models)
  111. - Fix ESpeakNG unsealed contents causing notarization failure
  112. - Fix GGUF vision model image handling and model deduplication
  113. ---
  114. ## Vesta 0.9.0 (Dec 10, 2025)
  115. ### New in 0.9.0
  116. - **Vision capabilities** with Qwen3-VL model (describe images, analyze screenshots)
  117. - **Continuity Camera** input (capture photos directly from iPhone/iPad)
  118. - **Code syntax highlighting** for 20+ programming languages
  119. - **Edit responses** inline after generation
  120. - **HTML preview** for rendered content
  121. - **Enhanced LaTeX** math rendering in blockquotes
  122. - Improved rendering engine with real-time code block highlighting
  123. ---
  124. ## Features
  125. - **Apple Intelligence** -- on-device AI via Foundation Models framework (always available)
  126. - **MLX Backend** -- Apple Silicon optimized inference with mlx-swift (Qwen3-VL vision models)
  127. - **llama.cpp Backend** -- GGUF model support with full Metal GPU acceleration and Jinja templates
  128. - **HuggingFace Explorer** -- cloud inference, image/video generation, transcription via 16+ providers
  129. - **External AI** -- connect to any OpenAI-compatible API server (LM Studio, Ollama, etc.)
  130. - **Vision** -- image understanding via Qwen3-VL (MLX, llama.cpp, or HuggingFace)
  131. - **Text-to-Speech** -- Kokoro, Marvis (with voice cloning), and Orpheus TTS engines
  132. - **Speech-to-Text** -- WhisperKit on-device transcription (Tiny through Large V3)
  133. - **MCP Server** -- 41+ tools for programmatic control, model management, and AI Sidekick integration
  134. - **GitHub Flavored Markdown** -- tables, task lists, strikethrough via remark/rehype pipeline
  135. - **LaTeX Math** -- inline and block math rendering with KaTeX
  136. - **Code Highlighting** -- 20+ languages with real-time streaming highlight
  137. - **Liquid Glass UI** -- native macOS Tahoe design
  138. - **App Sandbox** -- Developer ID signed and Apple notarized
  139. ## Verify
  140. ```bash
  141. # Check SHA256
  142. shasum -a 256 ~/Downloads/Vesta-*.dmg
  143. # Check code signature
  144. codesign --verify --deep --strict /Applications/Vesta.app
  145. # Check notarization
  146. spctl --assess --type execute /Applications/Vesta.app
  147. ```
  148. ## Requirements
  149. - **macOS 26.0** (Tahoe) or later
  150. - **Apple Silicon Mac** (M1/M2/M3/M4)
  151. - Microphone access for voice input and STT
  152. - Internet access for HuggingFace backend and model downloads (on-device backends work offline after model download)
  153. ## Security & Privacy
  154. - Signed with **Developer ID Application: Soprano Technologies Inc.**
  155. - **Notarized by Apple**
  156. - **App Sandbox enabled**
  157. - On-device backends (Apple Intelligence, MLX, llama.cpp) process everything locally -- no data sent to servers
  158. - HuggingFace and External AI backends require network access for inference
  159. - API tokens stored in macOS Keychain
  160. ## Related
  161. - **Source Code**: https://github.com/scouzi1966/vesta-mac
  162. - **Distribution**: https://github.com/scouzi1966/vesta-mac-dist (this repo)
  163. - **Website**: https://vesta-mac.pages.dev
  164. - **CLI Alternative**: https://github.com/scouzi1966/maclocal-api
  165. ## Support
  166. - [Report an Issue](https://github.com/scouzi1966/vesta-mac-dist/issues/new)
  167. - [Request a Feature](https://github.com/scouzi1966/vesta-mac-dist/issues/new)
  168. - [Browse Issues](https://github.com/scouzi1966/vesta-mac-dist/issues)
  169. ## License
  170. (c) 2025-2026 Soprano Technologies Inc. All rights reserved.
  171. ## Built With
  172. - **Apple Intelligence** -- Foundation Models framework
  173. - **MLX** -- mlx-swift + mlx-swift-lm for Apple Silicon inference
  174. - **llama.cpp** -- GGUF inference with Metal acceleration
  175. - **WhisperKit** -- CoreML-based Whisper speech-to-text
  176. - **mlx-audio** -- Kokoro/Marvis/Orpheus TTS
  177. - **SwiftUI** -- native macOS interface
  178. - **KaTeX** -- math rendering
  179. - **highlight.js** -- code syntax highlighting
  180. - **remark/rehype** -- markdown processing pipeline
  181. ---
  182. Built with automated distribution pipeline | Notarized and code-signed | Apple Silicon native
  183. ## Star History
  184. [![Star History Chart](https://api.star-history.com/svg?repos=scouzi1966/vesta-mac-dist&type=Date)](https://star-history.com/#scouzi1966/vesta-mac-dist&Date)

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.