ExecuTorch

ExecuTorch is Meta's production-grade framework for on-device AI inference, allowing PyTorch models to run natively on mobile phones, wearables, embedded systems, and AI PCs without cloud connectivity. Unlike conversion-based pipelines, ExecuTorch exports models directly from torch.export to a .pte binary format, retaining PyTorch semantics through the entire deployment stack. It reached general availability (v1.0) in October 2025 and is already in production across Meta's Ray-Ban Smart Glasses, Meta Quest headsets, and billions of on-device AI feature interactions on Instagram, WhatsApp, and Messenger.

General
GA date	18 Oct 2025 (v1.0)
Developer	Meta / PyTorch
Type	On-Device AI Inference Framework
License	BSD License
GitHub	pytorch/executorch
Documentation	docs.pytorch.org/executorch

Core Features

PyTorch-native export — models go from torch.export directly to .pte format with no ONNX or TFLite conversion step.
50 KB base runtime — minimal core footprint suitable for microcontrollers and embedded targets.
Ahead-of-time compilation — models compiled offline to .pte binaries, reducing on-device startup overhead.
Single-line backend switching — swap hardware accelerators (CPU, NPU, GPU) without rewriting model code.
Quantization tooling — INT8, INT4 (per-block), QAT+LoRA (QLoRA), and SpinQuant quantization via integrated PyTorch tools.
Selective operator builds — include only operators the model uses, minimizing binary size.
Multimodal support — composable backbone for LLMs, vision-language models, image segmentation, depth estimation, OCR, ASR, and object detection.
Hugging Face Optimum-ExecuTorch — over 80% of the most-downloaded edge-friendly models on Hugging Face run on ExecuTorch out of the box.

Supported Hardware Backends

Backend	Target	Status
XNNPACK + Arm KleidiAI	CPU (Android, iOS, Linux, AI PCs)	Stable
Apple Core ML	Apple silicon (iOS, macOS)	Stable
Qualcomm AI Engine / Hexagon NPU	Android (Qualcomm SoCs)	Stable
Arm Ethos-U NPU	Embedded / MCU	Stable
Vulkan GPU	Cross-platform GPU (Android, Linux)	Stable
Apple MPS (Metal Performance Shaders)	iOS / macOS GPU	Alpha
MediaTek NPU	Android (MediaTek SoCs)	Beta
Samsung Exynos NPU	Android (Samsung SoCs)	Alpha
Intel OpenVINO	AI PCs (Windows / Linux x86)	Alpha
CUDA	Linux / Windows GPU	Experimental

Hardware partners include Apple, Arm, Cadence, Intel, MediaTek, NXP Semiconductors, Qualcomm, and Samsung.

Performance (Llama 3.2 1B Quantized)

Device	Decode Speed	Prefill Speed
Samsung Galaxy S24+	>40 tokens/s	>350 tokens/s
OnePlus 12	50.2 tokens/s	260 tokens/s

Quantization reduces model size by ~52% (2.3 GiB to 1.1 GiB) and peak runtime memory by ~39%, with 2.5x average decode latency improvement over BF16 baseline.

Tools and Resources

GitHub — github.com/pytorch/executorch — source code, model export scripts, and backend integrations.
Documentation — docs.pytorch.org/executorch/stable — installation, export workflow, and backend guides.
PyPI — pip install executorch — Python package for model export and tooling.
Llama export script — export_llm command in the repo for exporting Llama variants to .pte format.
PyTorch blog — pytorch.org/blog/introducing-executorch-1-0 — GA announcement with performance benchmarks.

Ecosystem and Integrations

Powers on-device AI in Meta Ray-Ban Smart Glasses (live translation, visual captions, menu translation) and Oakley Meta Vanguard glasses (athletic performance insights).
Runs scene understanding, depth estimation, hand tracking, and persistent room memory on Meta Quest 3 / Quest 3S.
Llama 3.2 1B and 3B models were co-developed with Qualcomm and MediaTek for optimized Snapdragon deployment via ExecuTorch.
Backend available in Hugging Face Optimum-ExecuTorch for direct integration with the Hugging Face model hub.
Complements PyTorch Mobile for teams already in the PyTorch ecosystem, offering a significantly smaller runtime and better edge-hardware coverage.

Get started by cloning github.com/pytorch/executorch and following the quickstart guide, or install via pip install executorch for model export tooling.

Super Secure SYST v2.3 - 6-Gas AI Guardian

SUPER SECURE SPACE GUARDIAN - SIMULATOR TEORI UNTUK JUDGING DIGITHOOK ⚠️ INI SIMULATOR TEORI 01-02-03 BUAT JUDGING PAK - YANG AKTUAL ROBOTIK NDIR 4 PLUS THERMAL PLUS CCTV LAGI KITA RAKIT DI RUANG TEDUH - POWER WAJIB KABEL DATA WIFI/KABEL LAN ESP32 OTAK PLUS KONEKSI INCLUDE 01 THEORY: Fire risk theory: thermal anomaly >600C + HC >10% LEL = Elevated risk. NDIR 4-ch: 3.33um CH4/BBM/LPG/Ethanol 45C, 4.26um CO2, 4.64um CO, 3.91um Ref. Sekarang jamannya WiFi serba wireless, CCTV saja terhubung hanya WiFi walau mesti ada kelistrikan yang terus memberi power. Untuk mall/plaza/DPR, power WAJIB KABEL dan data bisa kabel LAN biar stabil. 02 OBSERVATION: Simulator dashboard F-21 HMDS 85% TRANSP di super-secure-syst-v23.streamlit.app - HUD GUDANG JAKARTA AREA C12. 30-sec DETECT / 270-sec STANDBY, sampling 5 menit, 90 hari circular log. Data gas di dashboard ini SIMULASI UI UNTUK DEMO, bukan NDIR aktual. 03 OPINION: Jika aktual robotik (ESP32 di ruang teduh) baca HC >10% LEL + thermal white hot 612C + O2 drop >0.5% = Confirmed heat + flammable vapor. Action: pompa 350mL ON, cut gas valve relay, buzzer+strobo, laser pointer, Evolus Workflow alert via AMD MI300X. ARSITEKTUR AKTUAL RUANG TEDUH: HIDUNG NDIR 4 (kabel power) -> OTAK+KONEKSI ESP32-S3 (WiFi include, power kabel) -> MATA MLX90640 Thermal 32x24 + IMX219 CCTV (power kabel, data WiFi/LAN) -> MONITOR APP PY (baca Serial COM3/MQTT, BUKAN random). GITHUB: Alchaliveret/super-secure-syst-v23 - README sudah ada disclaimer simulator teori. Alat aktual tidak di-deploy di cloud, diuji di lab ruang teduh untuk enterprise.

Aichaliveret - Smart Cotton SCM

Problem: Traditional garment factories in Indonesia still use handwritten notebooks for cotton fabric. 1 ball is 1201-1599 yards consisting of several rolls, cut to 180cm = 20-50 kodi only, 240cm = 20-50 kodi only, etc. This repeats daily/weekly/monthly. Manual counting is cheap (10 rupiah/paper) but causes repeated clerical work, mixed calculations across sizes, lost data, and no visibility for Direktur, Manufacture, and CMT Konveksi. Existing apps are complex, cost 200rb/month, and create phobia with too many tables. Solution: Aichaliveret v20 SCM Brandnew - a brandnew form.gle experience that is efficient and effective. Input data per ball only but continuous and repeating. Operator selects potongan size (180,240,270,300,340,Custom) - system isolates that size and never mixes calculations (jangan dijumlah campur aduk). Operator types Jumlah Yard 1201-1599 (validated), adds rolls via + button (Google Form style - Bisa otomatis tambah kotak diatas). System auto-calculates: Kodi = Yard*0.9144/(cm/100)/20 and validates 20-50 kodi saja per ball. Save creates "Potongan kain ukuran 180cm tersimpan" in its own table, separate from 240cm table. Target users & interaction: - Laman 1 Direktur: sees all - all sizes tables, total yard, WIP Jadi Kirim, transport supir. - Laman 2 Manufacture: operator gudang - input ball, roll, select potongan, save to Gudang/Mayan. - Laman 3 CMT: receives kirim, confirms Diterima, inputs Dijahit, auto Jadi Pakaian Jumlah dalam Kodi. - Supir: each ball is connected to transport data - nama, nomer KTP/SIM, no polisi, tujuan Sigading diketter, GPS Maps, Recant Klik Selesai, Pesan dan Saran. What makes it creative & unique: - Per-size isolation logic: 1 ball 1201-1599 yard = 20-50 kodi for THAT size only, not mixed. No other SCM does this. - Form.gle brandnew UI: single centered column 780px, white cards, purple top bar, + button to add box - no table phobia. - Efficient: localStorage auto-save like "All changes saved in Drive", works offline, deploy to Netlify jolly-cactus-5c95be.netlify.app. - Transparent formula: pcs = meter / (potongan/100), kodi = pcs/20, inline validation. Effectiveness: Reduces paper waste, replaces 200rb/month complex system with simple free static site, prevents mixing errors, gives Direktur full control, speeds up daily 20-50 kodi flow from Gudang to CMT with GPS-tracked supir. Tested with real handwritten diagram from factory. Tech: Single HTML file, vanilla JS, localStorage, SVG flowchart Direktur-Manufacture-Gudang-CMT.

THREENITY2CAMSCANNER - 7 DETIK LENSA INFRA

IR Face, Climate and Money Scanner 7 Detik V30 DUAL CAM is privacy-first early warning, not diagnosis. IR detects inflammation heat +0.5 to +1.5C. Fungus active border more hot, comedo more cold. Problem: 58% oily T-zone pores 0.3-0.6mm comedo fungus, food basi Quaker Oatmeal kemasan consumer goods not FMCG kasar kering kadaluarsa, IDR palsu fotocopy. Old PWA only front cam foto hitam bug. Solution 3WAYS 2CAM: 1. DERMA 15cm 7s front cam: pores comedo -0.2C fungus +0.8C acne +1.2C. Thermal color blue cold green normal red hot. Score + formula + doctor. 0-2s pos 2-5s scan 5-7s 500x observe. 2. CLIMATE rear cam 15cm: nasi uduk telor balado fast food ayam burger + QUAKER OATMEAL kemasan consumer goods. Deteksi kasar kering moisture loss kadaluarsa apek tengik red/yellow thermal. CCTV indoor HP jadi CCTV putar 180deg ruko 10-20 orang kantor 100-500 hilir mudik ukur humidity temp real-time ruangan kosong 10s timer standby. 3. MONEY rear cam 15cm aktif terus: IDR Bank Indonesia IR ink khusus asli serap IR hitam 0 palsu fotocopy pantul putih 255 uji fotocopy valid physics IR-Absorb not error. V30 Fix: Dual Cam front+back Switch Camera like Streamlit, fix foto hitam now color, auto switch by mode. 100% on-device Snapdragon NPU + Meta ExecuTorch. PWA magenta-torte-ir-7detik-3ways.netlify.app + Streamlit + GitHub + YouTube Shorts. Lablab infrared Derma Climate Money.

Fresno LUX: A Living City

Fresno LUX is a prototype for a living intelligence layer across an entire city. Fresno already contains enormous amounts of intelligence: residents know where problems exist, nonprofits know where help is needed, businesses know where opportunities exist, educators understand gaps in learning, researchers generate knowledge, and government maintains critical infrastructure and data. The problem is that much of this intelligence is fragmented across disconnected systems. Fresno LUX connects those pieces. The platform creates a shared city graph where people, organizations, resources, needs, projects, observations, experiments, and outcomes become connected nodes. Instead of simply collecting data, LUX continuously asks: What is happening? What do we know? What is missing? What can we test? What changed? A community problem can become an observable signal. That signal can become an experiment. The experiment produces evidence. Evidence changes the system's understanding. That updated understanding informs the next action. Observe → Connect → Understand → Act → Measure → Learn → Converge. The goal is to move Fresno from disconnected information toward collective intelligence—a system where knowledge can travel across boundaries and successful solutions can be discovered, tested, measured, and replicated. LUX is designed to make the invisible connections visible: the relationship between a neighborhood and its resources, between a student and educational opportunity, between a business and workforce capacity, between environmental conditions and community health, and between an idea and measurable real-world impact. This hackathon prototype demonstrates the beginning of a Living Fresno Graph: a city that can observe itself, learn from evidence, and coordinate toward better outcomes.

LifePilot — On-Device Wellness Co-Pilot

Every AI wellness app works the same way: your tasks, sleep, spending, and health data get shipped to someone else's servers before you get an answer. LifePilot removes that trade-off — every model runs on the device itself. LifePilot is a calm, offline-first wellness app with four features, each backed by its own on-device model: • Overwhelm Manager — describe what's overwhelming you; an on-device Llama 3.2 1B model breaks it into a concrete 5–8 step plan, personalized from on-device memory that never syncs anywhere. • Energy Planner — a trained time-series model predicts your energy curve for the day from recent sleep and activity. • Hydration Tracker — a trained regression model sets a personalized daily water target and explains why. • Expense Scanner — point the camera at a receipt; on-device OCR plus trained models extract the merchant, total, currency, and category. No photo or text ever uploaded. Every model is exported to ExecuTorch and runs directly on the phone's own chip. There is no server-side inference path — put the phone in airplane mode and everything still works identically. That's not a fallback; it's the only mode. Where AMD fits: the four trained models (energy, hydration, and two expense models) are trained and exported on AMD Instinct MI300X GPUs via ROCm, on AMD's ROCm cloud notebooks — real GPU compute producing the exact .pte files that ship in the app. All four features are built and running on a real Android device today, including the on-device Llama agent generating multi-step task breakdowns in airplane mode. Privacy isn't a marketing line here — it's the architecture.

Lantern: Model anything

Lantern is an AI-powered mobile application that transforms a standard smartphone into a real-time 3D scanner using entirely on-device inference. Users simply point their phone at an object and walk around it for approximately 30 seconds while Lantern estimates depth, tracks camera motion, reconstructs geometry, and generates a dimensionally accurate 3D mesh, all without sending data to the cloud. Built with ExecuTorch, Lantern demonstrates how edge AI can deliver fast, private, and offline 3D reconstruction on consumer smartphones. By eliminating the need for expensive LiDAR hardware, external computers, or cloud processing, Lantern makes high-quality 3D scanning more accessible to engineers, designers, makers, researchers, and students. The generated 3D models can be used for reverse engineering, CAD workflows, rapid prototyping, digital asset creation, manufacturing, and AR/VR applications. Lantern showcases the power of on-device AI by combining computer vision, monocular depth estimation, and real-time 3D reconstruction into a seamless mobile experience. Our goal is to make professional-quality 3D scanning available to anyone with a smartphone while preserving user privacy and enabling low-latency performance through efficient edge inference.