Apple Unveils M5: A Major Leap in AI Performance for Apple Silicon
Apple today introduced the M5, a new generation of silicon designed to accelerate on‑device AI workloads and deliver faster overall performance across its trio of devices—the 14-inch MacBook Pro, iPad Pro, and Apple Vision Pro. Built on third‑generation 3nm process technology, M5 combines a new 10‑core GPU architecture with Neural Accelerators in each core, a more capable CPU, a faster Neural Engine, and higher unified memory bandwidth to push AI and graphics to new heights. Pre-orders are open now.
“M5 marks the next big leap in AI performance for Apple silicon,” said Johny Srouji, Apple’s senior vice president of Hardware Technologies. “With Neural Accelerators in every GPU core, a faster CPU core, an improved Neural Engine, and greater memory bandwidth, M5 unlocks substantially stronger AI performance across MacBook Pro, iPad Pro, and Vision Pro.”
What’s new in M5
- AI-optimized GPU with Neural Accelerators in every core
- A 10‑core GPU featuring a dedicated Neural Accelerator per core enables GPU‑based AI workloads to run dramatically faster — over 4x the peak GPU compute versus M4, and more than 6x AI performance versus M1.
- Enhanced graphics capabilities and a third‑generation ray‑tracing engine deliver up to 45% graphics uplift compared with M4, with smoother visuals and faster rendering in demanding scenes.
- New shader cores, second‑generation dynamic caching, and integrated ray tracing help power immersive visuals in apps and games, including higher frame rates on Vision Pro with brighter, more detailed scenes.
- The architecture is designed to work hand‑in‑glove with Apple software frameworks like Core ML, Metal Performance Shaders, and Metal 4, so developers see immediate performance gains and can program the Neural Accelerators via Tensor APIs.
- CPU and memory for AI at scale
- The M5 pairs the world’s fastest performance core with up to 10 CPU cores (six efficiency cores and up to four performance cores), delivering up to 15% faster multithreaded performance than its predecessor.
- A faster 16‑core Neural Engine works alongside the CPU and GPU to maximize on‑device AI capabilities while keeping energy use remarkably efficient.
- On‑device AI tools, such as those in Apple Intelligence and the Photos 2D‑to‑3D capabilities, benefit from the boosted Neural Engine and the expanded unified memory pool.
- Unified memory bandwidth and capacity
- M5 increases unified memory bandwidth to 153 GB/s (roughly 30% higher than M4 and more than 2x M1), enabling larger AI models to run directly on the device.
- With up to 32 GB of unified memory, users can run demanding creative suites (e.g., Photoshop, Final Cut Pro) alongside large file transfers and background cloud uploads without compromising performance.
- Availability and compatible devices
- The M5 powers the refreshed 14-inch MacBook Pro, iPad Pro, and Apple Vision Pro, delivering improved AI responsiveness and graphics performance across these devices.
- All three devices are available for pre‑order, with software and developer tooling aligned to maximize on‑device inference and creative workflows.
- On‑device intelligence and ecosystem integration
- Apple Intelligence tools and the Foundation Models framework gain faster performance thanks to the upgraded Neural Engine and the broader memory pool.
- Developers can leverage built‑in Apple frameworks and APIs to accelerate AI workflows, or directly tap the Neural Accelerators for custom solutions.
- Environmental impact and performance efficiency
- Apple’s ongoing commitment to environmental goals continues with M5’s power efficiency, aiding the company’s 2030 carbon‑neutral ambitions by reducing overall energy use across devices during their lifetime.
Notes on testing and availability
- Apple conducted testing in September 2025 using a mix of preproduction and production systems to benchmark performance against prior generations. Results reflect selected industry standard benchmarks and configurations and may vary by workflow and software.
- Availability of features and language support varies by region and may require software updates or hardware configurations.
Selected image highlights (illustrative)
Stay tuned for more details on developer tools, software optimization, and timeline specifics as Apple expands AI capabilities across its ecosystem.