Machete DGX vs Pebble Mind

Machete DGX puts your NVIDIA DGX Spark in your pocket. Monitor, manage and experiment with your AI server from anywhere — live GPU dashboard, full SSH terminal, Docker management, local LLM chat and a built-in Hugging Face model hub, all in one app. LIVE GPU DASHBOARD • Real-time GPU, CPU, memory, disk and network metrics • Temperature and utilization charts at a glance • Catch problems before they cost you a training run SSH TERMINAL • Full SSH terminal with a real terminal emulator • Multiple sessions, hardware-keyboard friendly DOCKER MANAGEMENT • List, start, stop and inspect containers • Stream container logs in real time CHAT WITH YOUR OWN LLM • Talk directly to the LLM running locally on your DGX Spark • Your prompts stay on your hardware, not in someone else's cloud HUGGING FACE MODEL HUB • Search Hugging Face without leaving the app • Download models straight to your server • Launch them with fully editable runtime parameters COMMUNITY SOLUTIONS • Ready-to-run model recipes shared by the community • Star the best ones and run them in a tap • Publish your own configurations ALERTS & REPORTS • Push and e-mail alerts for temperature, utilization and job events • Daily e-mail reports with your server's health summary SCHEDULED BACKUPS • Back up to S3, Google Drive, Dropbox, Box, FTP or SCP • Set the schedule once, get notified about every run ONE-TAP SETUP • Install the on-server agent directly from the app — no manual configuration MAKE IT YOURS • Deep theming and customization • 9 interface languages • Settings sync across devices with a Supabase account PRIVACY FIRST • SSH credentials never leave your device • Your Hugging Face token is stored encrypted — in the device Keychain and in Supabase Vault • Direct device-to-server connections MACHETE PRO Monitoring your Spark is free. Machete Pro unlocks the power features: • Download and launch Hugging Face models on your Spark • SSH terminal and console access • Docker container management • 24/7 agent: alerts, scheduled backups and reports • Image generation and LLM chat • Multiple servers and the live training monitor Machete Pro is an auto-renewable subscription, offered monthly or yearly. Payment is charged to your App Store account at confirmation of purchase. The subscription renews automatically unless auto-renew is turned off at least 24 hours before the current period ends; the renewal is charged within 24 hours before the period ends. Manage or cancel anytime in your App Store account settings. Privacy Policy: https://dgx.down.pl/machete/privacy/ Terms of Use (EULA): https://dgx.down.pl/machete/terms/ Built for developers, ML engineers and homelab enthusiasts who want their DGX Spark within arm's reach at all times. Requires an NVIDIA DGX Spark (or compatible Linux host) reachable over SSH. This app is not affiliated with or endorsed by NVIDIA Corporation.

Your AI. Your models. Your device. Pebble Mind is a private, local-first AI workspace for running language and multimodal models on iPhone and iPad. Powered by llama.cpp, it brings local AI chat to your device without an account, API key, or remote inference service. Download a compatible model and keep chats and llama.cpp inference on device. PRIVATE BY DEFAULT Your conversations, attachments, settings, and downloaded models are stored locally. Prompts sent to downloaded llama.cpp models do not need to leave your device. Web Search is optional; when enabled, it can access current web information and show sources in your conversation. CHOOSE YOUR MODEL Browse text and multimodal models in one library. Check download size, device compatibility, memory requirements, and availability. Download GGUF models, switch between models, and manage storage in the app. MORE THAN TEXT Compatible multimodal models can work with photos, camera captures, PDFs, documents, and extracted text—all in the familiar chat experience. RECIPES FOR EVERYDAY TASKS Recipes turn personal information into practical next steps. Tailor a request, review results, and confirm before supported changes are made. Access is requested for the data source each recipe uses. • Plan what is next — Prepare for calendar events, schedule clashes, and travel days. • Explore nearby — Find restaurants, cafés, parks, museums, landmarks, and activities around your requested area. • Find patterns in Photos — Search objects and visual patterns, or review similar photos. Analysis runs on device; confirm before adding Favorites or deleting selected photos. • Reconnect with people — Find birthdays, possible duplicate contacts, and saved details; review and confirm merges. • Triage my reminders — Review due dates, lists, priorities, and undated tasks; choose reminders to complete or reschedule. • Understand my wellness — Explore recorded activity, sleep, cardio, vitals, mobility, body composition, nutrition, hydration, mindfulness, and audio exposure. Health data is read-only and not medical advice. • Explore my music taste — Find patterns in saved music by artist, genre, album, or release year; create playlists from tracks you select. • Plan around daylight — Check approximate sunrise and sunset times and prepare for outings. These are not live weather or safety forecasts. Recipe availability depends on your device. Some recipes require Pebble Mind Pro. BUILT FOR LOCAL AI • On-device llama.cpp inference • Streaming responses and text/multimodal models • Searchable conversations and model library • Optional web search with visible sources • Markdown responses; stop and retry generation • Per-model Temperature, Top-P, Top-K, context, and reply-length controls • Model download and storage management • Conversation sharing and export • Native, privacy-focused iOS experience PEBBLE MIND PRO Pebble Mind Pro is an optional one-time purchase, not a subscription. It unlocks custom model importing, expanded model management, advanced controls, generation metrics, data transfer, configurable search domains, context compaction, and additional Recipes. NETWORK ACCESS, WHEN YOU CHOOSE IT Internet access is used for model downloads, optional Web Search, and Explore nearby. Nearby searches use Apple Maps and share your query and approximate location with Apple. Model files may be downloaded from Hugging Face; web-search queries may be sent to external search services. Run AI locally. Choose your models. Keep control of your data.