NullMoth NVIDIA Driver Brings RTX 5060 Graphics to macOS Sequoia on Intel Hackintosh PCs

What was actually released​

VideoCardz reports that NullMoth has released a third-party NVIDIA graphics driver for macOS 15 Sequoia, reportedly enabling hardware-accelerated graphics on modern GeForce GPUs, including the RTX 5060. The developer claims that desktop rendering, Metal 3 applications and games run directly on NVIDIA hardware, not through software rendering. Those are the developer's claims. They are not independent benchmarks or certification.

The repository lists these Metal 3 features: argument buffers tier 2, ray tracing, mesh shaders, MPS and a MetalFX path. It also lists OpenGL through Apple's GL-on-Metal layer, OpenCL, Core Image and Core ML. The repository says application behavior depends on the card, the operating system version and the installed components. Treat the feature list as a statement of intent, not a guarantee that any given app or game works.

How the stack works​

This is a translation stack, not a port of NVIDIA's Windows driver. The documented chain is:

1. A Metal app talks to a Metal driver plugin.
2. A translator converts Apple's AIR shader format to SPIR-V.
3. Mesa's NVK Vulkan driver, with the NAK compiler, handles the Vulkan side.
4. Kernel extensions talk to the GPU.

On the kernel side, the project says it runs NVIDIA's own open GPU kernel modules (r610) under macOS. It uses NVIDIA's GSP firmware unmodified. It names four kernel extensions: NVRM, NVAccel, NVRMFB and NVRMAGDC. NVRMFB is the display framebuffer. NVAccel is the accelerator that WindowServer composites through. NVRMAGDC is a display-policy shim.

One of the most interesting parts is NVK. This is an open Vulkan driver originally built for Linux, and here it is bent into serving as the back end for Apple's graphics API. The shader translator is LGPL-3.0 and is based on an existing project called metal2vulkan.

"Listed" is not "tested"​

The driver's device table covers many cards. The installer carries 235 distinct display-card PCI device IDs from NVIDIA's 610.57.04 table. That covers GTX 16, RTX 20, 30, 40 and 50 series, TITAN RTX and some workstation cards.

The project is open about how far that coverage goes. A device-table match does not establish that a particular laptop's display routing, firmware, BAR placement, or macOS installation works. The validation picture is:

• Physically validated: the RTX 5060, on macOS 15.7.x and 15.8.1. VideoCardz says the full installation was validated on an RTX 5060 with an Intel Core Ultra 5 225F and a B860 motherboard.
• Reported working: the project's own card-support notes say RTX 5070 and 5080 are reported working. VideoCardz also cites Reddit users reporting success with the RTX 5070 Ti, RTX 3090 and RTX 3060. Those are community reports, not verified support.
• Reported trouble: release notes say that a 16 GB BAR window stopped some RTX 3060 and RTX 5080 cards during startup, so the driver now enlarges the window to at most 8 GB. Support for those cards is therefore still moving.

The README warns: this driver is new and may not work on every PC. Read "RTX 5060 runs macOS" as "one RTX 5060 system, plus community reports". It does not mean NVIDIA cards now work on macOS in general.

The price of admission: security and boot changes​

This is not a signed, consumer-grade installer. The documented OpenCore setup needs a long list of changes:

• SIP and boot arguments. The project uses csr-active-config set to <430A0000> (the tested value). Its boot-args include nvfb=1 nvaccel=1 nvfbheads=4 and AMFI-related arguments so WindowServer can load the driver bundle.
• Apple Secure Boot. The README sets SecureBootModel to Disabled, because Apple Secure Boot refuses kexts Apple did not sign. Release notes for 1.1.0 mention preserving signed recovery loading with SecureBootModel enabled, so this area is changing.
• BAR settings. The README gives ResizeGpuBars and ResizeAppleGpuBars values for the installed driver. The macOS installer needs the opposite values, because it has no NVIDIA driver and runs on the firmware's screen.
• A kernel block. The IONDRVSupport kext must be blocked, or the firmware framebuffer takes display index 0 from NVRMFB.
• BIOS and SMBIOS. Above 4G Decoding must be on and CSM off. A Mac model that runs macOS 15 with a discrete GPU is required, and iMacPro1,1 is the tested one.
• Extension approval. After a manual install, macOS prompts you to allow the extensions in System Settings → Privacy & Security, and you reboot again.

These changes lower macOS's defenses. They are fine for a hobbyist test machine. They are a poor fit for a machine holding sensitive work.

The Windows angle: the 1401 utility​

Windows users get a role here. VideoCardz notes that NullMoth's application called 1401 helps users prepare an OpenCore installation from Windows and configure the NVIDIA driver under macOS. The developer's Windows tool checks the PC, builds an OpenCore setup and writes a macOS installer USB drive. A separate Mac app installs the driver into an existing OpenCore setup.

The README says the Mac app does the following:

1. It finds the OpenCore that started the Mac.
2. It shows every change before making it.
3. It backs up the config and installs the driver.
4. It adds a "1401: Remove NVIDIA driver" entry to the OpenCore boot picker. Choosing that entry removes the driver at the next start and restores the previous setup.

Practical cautions from the documentation:

• Keep the OpenCore USB stick or disk that booted the Mac attached while the app runs, and for every restart afterward. Automatic copying to an internal disk is disabled.
• The first boot with the driver can pause for up to a minute at "PCI configuration end" while the GPU comes up.
• Back up your data and pick the macOS partition carefully. Do not wipe a disk that holds Windows.
• OpenCore's picker can boot Windows and Linux alongside macOS, according to the README's dual-boot notes.

To check a manual install after rebooting, the README suggests listing the loaded kernel extensions with kmutil showloaded --list-only | grep nullmoth, which should show 4 lines. You can also run system_profiler SPDisplaysDataType and look for your GeForce with Metal listed as supported.

Pace, licensing and the Tahoe question​

The project is moving fast. GitHub release notes show version 1.1.0 and 1.1.1 on October 8. Version 1.1.1 lists a known issue: with two displays, changing the refresh rate or resolution of one can make the other flash or go blank until you restart. Version numbers will have changed by the time you read this.

Licensing matters too. The README says the driver is free with source included, but the core components are under PolyForm Noncommercial 1.0.0. That means personal, research, educational and non-profit use is allowed, and commercial use is not.

For macOS 26 Tahoe, the README says full hardware and application qualification is still pending. It tells users not to treat the package contents as a verified upgrade path. So this is a Sequoia-only story for now.

How it compares with earlier efforts​

VideoCardz places this against two earlier RTX 5090 experiments on Apple Silicon:

• Tiny Corp's TinyGPU: an RTX 5090 connected to an M4 Pro Mac mini through Thunderbolt. It targeted AI and compute work, not Metal graphics or gaming.
• Scott Goldman's eGPU experiment: an RTX 5090 with an M4 MacBook Air, running Crysis and Cyberpunk 2077 through a Linux virtual machine using GPU passthrough, Proton and FEX.

NullMoth's approach differs in architecture and audience. It aims at direct macOS graphics integration, but only for Intel-based Macs and OpenCore PCs. That comparison comes from VideoCardz.

Analysis: who should care?​

The project is a notable technical feat, and it is open source, so others can inspect it. Its main value is for Hackintosh builders and people running older Intel Macs who want modern NVIDIA hardware in macOS. The limits are real:

• Intel only, and macOS 15 only.
• One physically validated GPU, so check the card-support notes before buying hardware for this.
• Reduced macOS security settings and a non-trivial OpenCore setup.
• Apple has shifted its platform to Apple Silicon. This project does not change that.
• Apple and NVIDIA have not endorsed it, and the sources show no sign they will.

If you try it, use a spare machine or a separate drive, keep your OpenCore recovery media handy, and treat any reported compatibility for non-5060 cards as unconfirmed.