Skip to main content
Main content
Image
pd27-blog-2026-pd27-header

What's new in Parallels Desktop 27

August 21, 2026

Text

The biggest changes in this release focus on workloads that place heavier demands on a virtual machine.

Parallels Desktop 27 introduces a Metal-based OpenGL 4.3 graphics driver. In testing conducted by Parallels, OpenGL workloads ran up to 2.6 times faster than with the previous driver. Testing with ArcGIS Pro showed gains of up to 35% in selected workflows and a reduction of up to 24% in overall workflow time.[1]

On Macs with Apple M4 chips or later, Parallels Desktop 27 also adds support for accelerated AI and machine-learning operations in Linux virtual machines. Tests on an M4 Pro Mac running Ubuntu measured matrix computations up to seven times faster and neural-network inference up to 1.75 times faster.[2]

For enterprise customers, Parallels Desktop 27 includes improvements to SSO, alongside continued investment in product security, third-party assurance, and platform certification.

Full support for macOS 27 Golden Gate

Parallels Desktop 27 is fully compatible with macOS 27 Golden Gate.

Customers can upgrade their Mac and continue using their Windows environment without reinstalling Parallels Desktop or rebuilding an existing virtual machine. Both Standard and Pro editions support macOS 27.

Windows applications can run alongside macOS, appear in the Dock, and open files stored on the Mac. There is no need to restart the Mac each time you move between operating systems.

Parallels Desktop is authorized by Microsoft to run Arm versions of Windows 11 in a virtual machine on Apple silicon Macs.

Intel Support

Parallels Desktop 27 is available exclusively for Apple Silicon Macs. Parallels Desktop 26 and older versions are still available for Intel-based Macs. For the latest lifecycle information, please refer to the knowledge base article.

Parallels Desktop Standard: for everyday Windows applications

Standard is the best fit for people who need dependable access to Windows software but do not need advanced development or high-performance graphics features.

A student might use it for a Windows-only course application. An accountant may need Microsoft Access or specialist financial software. A small-business owner may rely on an older line-of-business tool that has no Mac version.

With Standard, these applications run on the same Mac as the rest of your work. You can open Windows files without moving to another computer and use Windows and Mac applications side by side.

Standard is suited to:

  • Microsoft Office applications, including Access
  • Financial, educational, and business software built for Windows
  • Windows-only productivity tools
  • Occasional or frequent Windows use on a Mac
  • Windows or Linux virtual machines with standard resource requirements

For customers whose main goal is to run familiar Windows applications, Standard provides the core Parallels Desktop experience without the additional technical controls included in Pro.

Parallels Desktop Pro: for development and demanding applications

Pro includes the features in Standard and adds higher virtual hardware limits, advanced networking, command-line tools, automation options, and support for workloads that depend heavily on graphics or performance.

It is intended for developers, engineers, GIS specialists, data professionals, designers, and other customers who use a virtual machine as part of their daily technical work, often running multiple development tools, datasets, and applications side by side.

Faster OpenGL graphics

Many GIS, CAD, engineering, and scientific applications use OpenGL to draw maps, models, terrain, and other complex visual data.

Parallels Desktop 27 Pro uses Apple's Metal graphics technology to support OpenGL 4.3 in Windows. Parallels testing measured performance improvements of up to 2.6 times over the previous graphics driver.[1]

ArcGIS Pro showed particularly useful gains. Selected tests ran up to 35% faster, while the time required to complete the full tested workflow fell by as much as 24%. [2]

Actual results depend on the Mac, the virtual machine configuration, the project, and the operations being performed. The figures reflect measured improvements in specific tests rather than a guarantee for every task.

Faster AI processing in Linux virtual machines

Developers increasingly use Linux virtual machines to test local models, build AI-enabled applications, or work with tools such as Ollama and llama.cpp.

On Apple M4 Macs and later, Parallels Desktop 27 Pro supports Scalable Matrix Extension instructions for compatible Linux workloads. In an Ubuntu virtual machine running on an M4 Pro Mac, Parallels measured matrix operations at up to seven times the previous speed and neural-network inference at up to 1.75 times faster. [3]

The benefit varies by model and software stack. Workloads that make substantial use of supported matrix operations are likely to see the clearest gains.

More control over virtual machines

Pro also gives technical users more room to configure and manage their environments. Depending on the Mac and product limits, customers can assign additional CPU cores and memory, manage several virtual machines, create snapshots, use advanced network configurations, and automate tasks through command-line tools.

These controls are useful when testing software across operating systems, maintaining isolated development environments, or reproducing a customer configuration without changing the host Mac.

Standard or Pro: which edition should you choose?

Choose Standard when Windows supports part of your day. It is designed for Office applications, education, financial software, and other general Windows programs.

Choose Pro when the virtual machine is itself a work environment. The additional graphics support, resource controls, developer tools, and M4 AI capabilities are intended for software development, ArcGIS Pro, engineering applications, Linux-based AI work, and multi-VM testing.

CapabilityStandardPro
Run Windows 11 on MacYesYes
Run everyday Windows applicationsYesYes
Run Linux virtual machinesYesYes
macOS Golden Gate supportYesYes
Metal-based OpenGL 4.3 driverNoYes
ArcGIS Pro performance improvementsNoYes
M4 AI and machine-learning acceleration in LinuxNoYes
Higher virtual hardware limitsNoYes
Advanced networking and command-line toolsNoYes
Multi-VM development and testing featuresLimitedYes

Parallels Desktop 27 supports Apple silicon Macs from M1 through M5. The AI acceleration described in this article requires Apple M4 or newer. If you're an individual developer, engineer, designer, or other technical professional, Parallels Desktop Pro Edition is the right place to start. Organizations deploying Parallels across multiple Macs should consider Parallels Desktop Enterprise Edition.

What's new for enterprise customers?

Parallels Desktop 27 continues to address the security and infrastructure requirements of organizations that deploy and manage virtual machines at scale. This release includes improvements to SSO, alongside continued investment in third-party assurance, product security testing, and platform certification.

SSO improvements

Parallels Desktop 27 includes improvements to SSO, helping organizations maintain a consistent authentication experience as part of their existing identity and access infrastructure.

A native interface for DevOps management

The latest update to the Parallels Desktop DevOps Service adds a native interface for viewing and managing hosts, virtual machines, and automation workflows.

Teams that prefer a graphical view can perform common management tasks without relying entirely on command-line tools. Command-line access remains available for scripts and repeatable automation. Jenkins dynamic provisioning can create temporary build or test environments when a job needs them, then release those resources when the job is complete. This gives development teams a way to add Mac-hosted virtual machines to existing CI/CD processes.

Renewed SOC 2 compliance

Parallels Desktop has renewed its SOC 2 compliance, providing organizations with continued third-party assurance around the controls and processes supporting our products and services. See our Trust Center for the full report.

Ongoing product security testing

Parallels Desktop undergoes penetration testing by a third-party vendor as part of our product security practices, helping us identify and address potential vulnerabilities as the product evolves.

Arm SystemReady certification

Parallels Desktop 26.4 achieved Arm SystemReady certification, demonstrating alignment with Arm’s architectural standards. Certification work for Parallels Desktop 27 is now underway.

Ready to try these enhancements in your organization? Start your free 30-day trial today.

Frequently asked questions

Does Parallels Desktop 27 support macOS 27 Golden Gate?

Yes. Both Standard and Pro editions fully support macOS 27 Golden Gate, so you can upgrade your Mac and continue using your existing Windows virtual machine.

What is the difference between Parallels Desktop Standard and Pro?

Standard is designed for everyday Windows applications, including Microsoft Office, financial software, educational tools, and other productivity apps.

Pro includes everything in Standard, plus the Metal-based OpenGL 4.3 graphics driver, AI and machine-learning acceleration for supported Linux workloads on Apple M4 Macs and later, expanded virtual hardware limits, advanced networking, and developer tools.

Which edition should I choose?

Choose Standard if you primarily use Windows for productivity, business, education, or other everyday applications.

Choose Pro if your work involves software development, engineering, GIS, CAD, multiple virtual machines, or local AI workloads that benefit from additional performance and virtualization features.

How much faster is Parallels Desktop 27 Pro?

Performance improvements depend on the workload.

Parallels testing showed OpenGL workloads running up to 2.6 times faster than the previous release. ArcGIS Pro workflows ran up to 35% faster, with overall workflow times reduced by up to 24%. Linux AI workloads on Apple M4 Macs achieved up to 7 times faster matrix computations and up to 1.75 times faster neural-network inference.[1][2][3]

Which Macs support the new AI acceleration features?

Parallels Desktop 27 runs on Apple silicon Macs from M1 through M5. The AI and machine-learning acceleration introduced in Pro requires an Apple M4 or newer.

Do Business and Enterprise editions include these improvements?

Yes. Business and Enterprise editions build on the capabilities in Parallels Desktop Pro while adding centralized deployment, security, licensing, and management features for organizations. Parallels Desktop 27 also includes improvements to SSO. Parallels has renewed SOC 2 compliance and continues third-party penetration testing. Arm SystemReady certification work for Parallels Desktop 27 is underway.

Find the edition that matches your work

For Office, Access, business applications, and other everyday Windows software, start with Parallels Desktop Standard.

Choose Pro when you need OpenGL 4.3, higher virtual hardware limits, development and networking tools, ArcGIS Pro improvements, or accelerated Linux AI workloads on a supported Apple silicon Mac.

Organizations planning a managed rollout can speak with the Parallels sales team about Business and Enterprise licensing, deployment, SSO, security, management and DevOps requirements.

Compare Parallels Desktop editions
Start a Parallels Desktop trial
Discuss an enterprise deployment

Performance disclosures

[1] Performance measurements were conducted by Parallels using a prerelease version of Parallels Desktop 27. Results vary based on workload, system configuration, virtual-machine settings, and other factors.

[2] ArcGIS Pro testing measured up to 35% faster performance in selected tasks and up to a 24% reduction in overall workflow time with the new Metal-based OpenGL 4.3 driver.

[3] AI and machine-learning tests were conducted in an Ubuntu Linux virtual machine on an Apple M4 Pro Mac. Results included matrix computations up to seven times faster and neural-network inference up to 1.75 times faster. The feature requires Apple M4 hardware or newer.