# Neural Engine

> Mediated Wiki article. Canonical URL: https://mediated.wiki/source/Neural_Engine
> Markdown URL: https://mediated.wiki/source/Neural_Engine.md
> Source: https://en.wikipedia.org/wiki/Neural_Engine
> Source revision: 1326235840
> License: Creative Commons Attribution-ShareAlike 4.0 International (https://creativecommons.org/licenses/by-sa/4.0/)

AI accelerator series by Apple Inc.

**Neural Engine** is a series of [AI accelerators](/source/AI_accelerator) designed for machine learning by [Apple](/source/Apple_Inc.). Neural Engine was first introduced with the [A11 Bionic](/source/Apple_A11) [system-on-a-chip](/source/System-on-a-chip) (SoC), used in the [iPhone 8](/source/IPhone_8), [iPhone 8 Plus](/source/IPhone_8_Plus) and [iPhone X](/source/IPhone_X) from 2017.[1] In 2020, Apple introduced its [M1](/source/Apple_M1) processor for its Mac computers[2] which also used a Neural Engine. Every A-series and M-series processor since 2017 has included a Neural Engine.[3]

Apple services such as its [Siri](/source/Siri) virtual assistant, [Face ID](/source/Face_ID) facial recognition and [Apple Intelligence](/source/Apple_Intelligence) AI services are powered by the Neural Engine,[4] and since this is handled on-device, user data is secure.[5]

## Applications

The Neural Engine is used for real-time AI-driven applications such as [Face ID](/source/Face_ID), [Siri](/source/Siri), and [augmented reality](/source/Augmented_reality) (AR). It also handles [computational photography](/source/Computational_photography) features, including Smart HDR and Night Mode, by processing large amounts of sensor data for real-time image enhancements.[4]

In 2024, Apple also introduced its [Apple Intelligence](/source/Apple_Intelligence) AI suite to [iPhone](/source/IPhone), [iPad](/source/IPad) and [Mac](/source/Mac_(computer)), which included features like improvements to Siri, creating images with '[Image Playground](/source/Apple_Intelligence#Image_Playground)' and proofreading and correcting text with '[Writing Tools](/source/Apple_Intelligence#Writing_Tools)'.[6]

## Energy efficiency and privacy

The Neural Engine also provides high energy efficiency, allowing real-time AI tasks to be performed with minimal battery consumption.[6] Its on-device processing ensures that sensitive tasks such as [facial recognition](/source/Facial_recognition_system) and [voice commands](/source/Voice_command) are handled locally, enhancing privacy by keeping user data secure.[5]

## Developer tools

The Neural Engine is fully integrated with Apple's [Core ML](/source/Core_ML) framework, which allows developers to run machine learning models on-device. This integration supports applications like [object recognition](/source/Object_recognition), [natural language processing](/source/Natural_language_processing), and gesture detection.[7]

## Performance

Apple has stated the Neural Engine in the [M4](/source/Apple_M4) can perform 38 trillion operations per second (TOPS), an improvement over the 18 TOPS in the [M3](/source/Apple_M3).[8]

## References

1. **[^](#cite_ref-1)** Butts, Jeff (2023-02-16). ["What Is the Apple Neural Engine and What Does It Do?"](https://www.macobserver.com/tips/deep-dive/what-is-apple-neural-engine/). *The Mac Observer*. Retrieved 2023-12-26.

1. **[^](#cite_ref-2)** ["Apple unleashes M1"](https://www.apple.com/newsroom/2020/11/apple-unleashes-m1/). *Apple Newsroom* (Press release). Retrieved 2023-12-26.

1. **[^](#cite_ref-3)** ["Apple's Neural Engine: How AI Works Inside A and M Chips - victorycomputer.pk"](https://www.victorycomputer.pk/apples-neural-engine-how-ai-works-inside-a-and-m-chips). *Victory Computer*. 27 August 2025. Retrieved 28 August 2025.

1. ^ [***a***](#cite_ref-neaib_4-0) [***b***](#cite_ref-neaib_4-1) ["Neural Engine AI Basics"](https://www.aionlinecourse.com/ai-basics/neural-engine). *AI Online Course*. Retrieved 26 December 2023.

1. ^ [***a***](#cite_ref-npu_5-0) [***b***](#cite_ref-npu_5-1) ["NPUs: Fueling the future of AI in the Digital Workplace - Atos"](https://atos.net/en/blog/npus-fueling-the-future-of-ai-in-the-digital-workplace). *Atos*. Retrieved 28 August 2025.

1. ^ [***a***](#cite_ref-idb_6-0) [***b***](#cite_ref-idb_6-1) ["Apple Intelligence system requirements: Is your device compatible?"](https://www.idownloadblog.com/2024/06/19/apple-intelligence-system-requirements-device-compatibility-list-iphone-ipad-mac-ios-ipados-18-macos-sequoia). *iDownloadBlog*. 19 June 2024. Retrieved 28 August 2025.

1. **[^](#cite_ref-7)** ["Does my Core ML model run on Apple's Neural Engine? - Fritz ai"](https://fritz.ai/does-my-core-ml-model-run-on-apples-neural-engine/). *Fritz.ai*. September 21, 2023. Retrieved September 21, 2025.

1. **[^](#cite_ref-8)** Andrew Cunningham (7 May 2024). ["Apple announces M4 with more CPU cores and AI focus, just months after M3"](https://arstechnica.com/apple/2024/05/apple-announces-m4-with-more-cpu-cores-and-ai-focus-just-months-after-m3/). *ArsTechnica*. Retrieved 8 May 2024. Apple says the M4 runs up to 38 trillion operations per second (TOPS) […] The M3's Neural Engine is only capable of 18 TOPS

---
Adapted from the Wikipedia article [Neural Engine](https://en.wikipedia.org/wiki/Neural_Engine) by Wikipedia contributors ([contributor history](https://en.wikipedia.org/wiki/Neural_Engine?action=history)). Available under [Creative Commons Attribution-ShareAlike 4.0 International](https://creativecommons.org/licenses/by-sa/4.0/). Changes may have been made.