PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteShort verdict: The Radeon RX 9060 XT is workload-dependent. It trails NVIDIA’s GeForce RTX 5060 by a substantial margin in HotHardware’s Procyon text-generation test, comes within about 10% of it in the FP16 Stable Diffusion XL test, and falls well behind an older RTX 3070 Ti in Blender 4.3. Its strongest showing here is Blackmagic RAW decoding, where it beats the RTX 4060 Ti in the tested 12:1 modes. These results do not establish one universal winner for AI, rendering, video or gaming.
What HotHardware tested
HotHardware published these AI and creator results on June 4, 2025 using a Gigabyte Radeon RX 9060 XT Gaming OC with 16GB of memory. The test system used an AMD Ryzen 7 9800X3D, 32GB of G.SKILL DDR5-6000 memory, an MSI X870E Carbon WiFi motherboard, Windows 11 Pro 24H2, and Resizable BAR with the memory’s EXPO profile enabled. AMD Radeon driver 25.5.1 and NVIDIA drivers 572.50/575.94 were used.
As an Amazon Associate I earn from qualifying purchases.
The comparison set included Radeon RX 7600 XT (16GB), RX 7700 XT, RX 9070, GeForce RTX 3070 Ti, RTX 4070, RTX 5070 and GeForce RTX 5060 Ti, along with the RTX 5060 in the relevant charts. The page covered MLPerf Client, Geekbench AI, Procyon AI, Blender 4.3 and Blackmagic RAW Speed Test 4.3.1. The findings below are qualitative because the readable source text does not provide the chart numbers.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →AI benchmark results
MLPerf Client inference
The RX 9060 XT delivered a wide improvement over the RX 7600 XT, but it did not catch the RX 7700 XT and remained behind the rest of the listed group by a considerable margin. MLPerf Client is an MLCommons benchmark for inference on client PCs, and its scores can reflect hardware and software optimizations allowed under the benchmark’s accuracy rules. Treat the result as a measurement of this approved test configuration, not a guarantee for every AI framework.
#1 Best Overall
- OC mode (GPU Tweak III): up to 3330 MHz (Boost Clock)/up to 2760 MHz (Game Clock) Default mode: up to 3310 MHz (Boost Clock)/up to 2740 MHz (Game Clock)
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- Dual-ball fan bearings last up to twice as long as standard conventional sleeve bearings designs
- 0dB technology lets you enjoy light gaming in relative silence
- Dual BIOS switch lets you toggle between Quiet and Performance BIOS profiles
Geekbench AI precision modes
On Geekbench AI, the card sat around the middle of the pack in the half-precision result. It performed better relative to the field in single-precision and quantized tests. Those are different numerical modes with different software paths, so one score should not be used as a single “AI performance” rating.
Procyon text generation
In Procyon’s text-generation workload, the RX 9060 XT landed between the RX 7700 XT and RX 7600 XT and trailed the GeForce RTX 5060 by a relatively large margin. HotHardware ran this test with ONNX and DirectML rather than TensorRT. The final Llama 2 run omitted 8GB GeForce cards because they could not complete it with their available memory.
That memory behavior matters when comparing cards: an 8GB model may run a smaller prompt or model yet fail a larger one, while a nominally slower card with 16GB can complete the workload.
Procyon Stable Diffusion XL
In the demanding FP16 Stable Diffusion XL image-generation test, the RX 9060 XT outperformed the RX 7700 XT and was about 10% behind the RTX 5060, according to HotHardware. The comparison used ONNX; the reviewer used it as a point of comparison for RTX 40-series cards because TensorRT had not yet been updated for Blackwell. Results therefore describe this backend and software state, not every Stable Diffusion installation.
Rank #2
- Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- 0dB technology lets you enjoy light gaming in relative silence
- Dual BIOS switch lets you toggle between Quiet and Performance BIOS profiles
- Dual ball fan bearings last up to twice as long as sleeve bearing designs
3D rendering in Blender
Blender 4.3’s three default model tests placed the RX 9060 XT between the RX 7700 XT and RX 7600 XT. It was well behind the older GeForce RTX 3070 Ti. Blender performance can change substantially with the render engine, OptiX or HIP backend, scene complexity and driver version, so this result should not be generalized to every Blender project.
Video decoding with Blackmagic RAW
The RX 9060 XT’s best showing on this page came in Blackmagic RAW Speed Test 4.3.1. It held a large advantage over the RX 7600 XT and beat the GeForce RTX 4060 Ti in the tested 12:1 modes at the reported 4K and 8K resolutions.
These are GPU-decoding results for Blackmagic RAW at specified resolutions and compression settings. They do not represent export speed, effects processing, timeline performance or every codec in DaVinci Resolve and other editing applications.
Results by workload at a glance
| Workload | RX 9060 XT result | Important test qualification |
|---|---|---|
| MLPerf Client | Far ahead of RX 7600 XT; behind RX 7700 XT and much of the field | Client inference benchmark; permitted software optimizations can affect scores |
| Geekbench AI | Mid-pack in FP16; relatively better in FP32 and quantized tests | Precision modes are separate measurements |
| Procyon text generation | Between RX 7700 XT and RX 7600 XT; well behind RTX 5060 | ONNX with DirectML; 8GB GeForce cards failed the final Llama 2 run for insufficient memory |
| Procyon Stable Diffusion XL | Ahead of RX 7700 XT; about 10% behind RTX 5060 | FP16 test using ONNX; TensorRT was not yet updated for Blackwell |
| Blender 4.3 | Between RX 7700 XT and RX 7600 XT; well behind RTX 3070 Ti | Three default model tests |
| Blackmagic RAW | Large lead over RX 7600 XT; ahead of RTX 4060 Ti in 12:1 tests | GPU decoding at the tested 4K/8K resolutions and compression settings |
RX 9060 XT model and memory details
AMD announced two versions in May 2025: an 8GB model with a $299 USD launch suggested price and 150W starting total board power, and a 16GB model at $349 USD with 160W starting total board power. Those are launch figures, not a current street-price check.
Rank #3
- "OC mode (GPU Tweak III): up to 3250 MHz (Boost Clock)/up to 2640 MHz (Game Clock) Default mode: up to 3230 MHz (Boost Clock)/up to 2620 MHz (Game Clock)"
- Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- 0dB technology lets you enjoy light gaming in relative silence
- Dual BIOS switch lets you toggle between Quiet and Performance BIOS profiles
Gigabyte’s official “Radeon RX 9060 XT GAMING 8G” page describes an 8GB card with a 128-bit memory interface, boost clock up to 3130MHz and a 2530MHz game clock. It lists WINDFORCE cooling, Hawk fans, RGB lighting, reinforced construction and a 3D Active Fan mode that stops the fans under low load or low-power gaming conditions. These are manufacturer specifications and feature descriptions, not independent measurements from HotHardware’s benchmark page.
Do not assume that the Gigabyte GAMING 8G product page is the same board as HotHardware’s Gaming OC review sample. The review sample had 16GB, while Gigabyte’s cited page is an 8GB model; AMD sells both capacities. Verify the exact card name and VRAM before buying or comparing results.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.RX 9060 XT versus RTX 5060 for AI and creator work
The answer depends on the application. In HotHardware’s Procyon text-generation test, the RTX 5060 had a clear advantage. In Stable Diffusion XL FP16, the RX 9060 XT was approximately 10% slower. The review does not establish an RTX 5060 advantage in Blender or Blackmagic RAW because those comparisons and methods differ; its Blender finding instead places the RX 9060 XT below an RTX 3070 Ti, while its Blackmagic result beats an RTX 4060 Ti in the tested 12:1 modes.
Recommended Free Tools
For a purchase decision, compare the exact application, model or scene, precision, resolution, VRAM capacity, driver, and acceleration backend. Also compare current prices for the exact 8GB or 16GB card; the historical launch suggestions cannot determine today’s value.
How to use these results when choosing a card
- Local language models: prioritize the model’s memory requirement and framework support. The failed 8GB Llama 2 runs show why capacity can matter as much as throughput.
- Image generation: check whether your software uses DirectML, ONNX, HIP, Vulkan, CUDA or TensorRT. The Stable Diffusion comparison used ONNX.
- Blender: identify the render engine and plug-ins before relying on a GPU ranking. The cited Blender result covers three default Blender 4.3 models.
- Blackmagic workflows: distinguish RAW decoding from editing, effects and export. The favorable result applies to the tested Blackmagic RAW 12:1 modes.
- Mixed use: weigh the workload you run most often rather than averaging unlike benchmarks into one score.
Gaming context is separate
WIRED’s separate gaming review reported that the RX 9060 XT traded blows with the RTX 5060 at 1080p and moved ahead in its 1440p results, with performance varying by game. That gaming comparison should not be applied to the AI, Blender or video-decoding results above.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
