(home)=
ExecuTorch is PyTorch's open source export and runtime stack for running AI locally on phones, wearables, laptops, browsers, embedded systems, and microcontrollers.
Match the documentation version to your installed ExecuTorch release using the
version selector. Examples on the `main` site may require a nightly package or
a source checkout.
::::{grid} 1 2 2 2 :gutter: 3
:::{grid-item-card} Run your first model :link: getting-started :link-type: doc
Install ExecuTorch, export MobileNet V2 to XNNPACK, and execute the resulting
.pte program on the host.
+++ Open the tutorial → :::
:::{grid-item-card} Choose a path :link: user-pathways :link-type: doc
Route to the right guide by model, workload, target platform, experience level, or role.
+++ Open the decision guide → :::
::::
- PyTorch-native deployment: Capture models with
torch.export, lower them for a chosen target, and retain PyTorch program metadata for debugging. - Target-specific acceleration: Delegate supported graph regions to CPU, GPU, NPU, and DSP backends; unpartitioned regions run with the kernels included in the runtime.
- A portable, right-sized runtime: Integrate
.pteprograms through C++, Python, Java/Kotlin, Objective-C/Swift, or JavaScript, and include only the operators and backends the application needs.
::::{grid} 1 1 3 3 :gutter: 2 :class-container: success-showcase
:::{grid-item-card} Billions of people :class-header: bg-primary text-white :class-body: text-center
Production features across Instagram, WhatsApp, Messenger, and Facebook run on-device with ExecuTorch.
Read the Meta engineering story → :::
:::{grid-item-card} Shipping voice transcription :class-header: bg-primary text-white :class-body: text-center
LM Studio uses ExecuTorch and Parakeet TDT for local transcription on macOS and Windows.
Read the production case study → :::
:::{grid-item-card} Up to 2× CPU throughput :class-header: bg-primary text-white :class-body: text-center
Liquid AI reports the gain against selected similarly sized models, plus lower memory use, on its tested laptop and mobile CPUs.
Read Liquid AI's case study → ::: ::::
{doc}Explore all deployments, integrations, and showcases → <success-stories>
::::{grid} 1 2 3 3 :gutter: 2
:::{grid-item-card} Core concepts :link: intro-section :link-type: doc
Architecture, export and runtime concepts, and the .pte program format.
:::
:::{grid-item-card} Export a model :link: using-executorch-export :link-type: doc
Capture, lower, quantize, and validate a PyTorch model for a chosen target. :::
:::{grid-item-card} Advanced optimization :link: advanced-topics-section :link-type: doc
Quantization, memory planning, custom operators, passes, and backends. :::
:::{grid-item-card} Deploy by platform :link: edge-platforms-section :link-type: doc
Android, iOS, desktop, and embedded integration guides. :::
:::{grid-item-card} Choose a backend :link: backends-section :link-type: doc
Compare CPU, GPU, NPU, and DSP acceleration paths for target hardware. :::
:::{grid-item-card} Work with LLMs :link: llm/working-with-llms :link-type: doc
Export, optimize, and deploy text and multimodal generation models. :::
:::{grid-item-card} Runtime and LLM APIs :link: api-section :link-type: doc
Find C++, Python, Java/Kotlin, Objective-C/Swift, and JavaScript APIs. :::
:::{grid-item-card} Optimize and debug :link: tools-section :link-type: doc
Profile execution and inspect programs with ETDump, ETRecord, and numeric debugging. :::
:::{grid-item-card} Get help and contribute :link: support-section :link-type: doc
Troubleshooting, FAQs, issue reporting, and contribution guidance. :::
::::
:hidden:
:maxdepth: 1
intro-section
quick-start-section
user-pathways
success-stories
edge-platforms-section
backends-section
llm/working-with-llms
advanced-topics-section
tools-section
api-section
support-section