The AMD Radeon RX 9070 GRE GPU finally arrives for the global audience (it was initially a China-only SKU), slotting in between the mainstream RX 9060 XT 16GB GPU and the higher-end RX 9070 (non-XT) GPU.

On 1 June 2026, AMD officially launched its Radeon RX 9070 GRE GPU for the global market - for those of you who are unaware, the RX 9070 GRE was first sighted back in April/May 2025, where it was a China-exclusive SKU. Close to a year later, AMD's finally made the decision to do a global launch for the RX 9070 GRE, amidst a dearth of new global GPU launches caused partially by the AI datacenter boom, which has robbed supply of VRAM for new graphics cards.

Built on the RDNA 4 architecture, the RX 9070 GRE utilises the "Navi 48 XL" GPU, with 3072 stream processors and a boost clock of 2.79 GHz, slotting it between AMD's mainstream RX 9060 XT 16GB GPU and the higher-end RX 9070 (non-XT) GPU. Paired with 12GB of GDDR6 VRAM, AMD's marketing spiel positions the RX 9070 GRE against NVIDIA's GeForce RTX 5060 Ti 16GB with a MSRP of US$549 (for context, the 5060 Ti 16GB had a launch MSRP of US$429, but we're instead seeing actual market prices of around US$569 due to several inflationary factors).

For this review, ALKtech unboxes and tests a RX 9070 GRE graphics card by XFX - the XFX Swift Radeon RX 9070 GRE Triple Fan Gaming Edition (RX-97GRE2 V1.0). Featuring a triple-fan cooler design, the Swift is XFX's value-oriented product in its line-up of graphics cards, where the card doesn't come with any RGB bling and simply gets the job done without any fuss. Read on to see how this card performs with a small selection of gaming benchmark tests at the 1440p and 1080p resolution, when paired with AMD's Ryzen 7 7800X3D gaming processor - we've also got a chart to show how it stacks up against the RTX 5060 Ti 16GB at both 1440p and 1080p as well!

Summary of Specifications (RX-97GRE2 V1.0)

• GPU: AMD Radeon RX 9070 GRE
• Memory: 12GB GDDR6 RAM (192-bit, 18 Gbps)
• Core Clocks: up to 3290 MHz (boost clock), up to 2700 MHz (game clock)
• Bus Type: PCIe 5.0 x16
• Outputs: 3x DisplayPort 2.1a, 1x HDMI 2.1b
• External Power Required: Yes, 1x 8-pin PCIe
• Recommended PSU Wattage: 650W and above
• Official XFX product page: http://xfxforce.com/shop/xfx-swift-amd-radeon-rx-9070gre-triple-fan-gaming-edition

Packaging & Accessories

The XFX Swift Radeon RX 9070 GRE Triple Fan 12GB GDDR6 ships in a rectangular box with portrait-style graphics as opposed to the usual landscape style of layout, which definitely helps it to stand out, packaging wise.

We don't get much in terms of accessories, where the card ships with a large rectangular leaflet with QR codes for regulatory information and a quick installation guide,

Aesthetics & Physical Features

The Swift Radeon RX 9070 GRE Triple Fan Gaming Edition 12GB GDDR6 has a stealthy all-black aesthetic that is devoid of any RGB lighting effects, where it has a plastic fan shroud housing three 90mm fans and a metal backplate.

Given the flak that the 12VHPWR/12V-2x6 connector's gotten over the past few months, with stories of end-users with burnt connectors aplenty, it's great to see that the 9070 GRE utilises the traditional 8-pin PCIe power connector, where we've got two of them, to be precise.

An immediate plus point over NVIDIA's GeForce RTX 5060 Ti is the fact that AMD's opted to go with a full-fat PCIe 5.0 x16 connection for the RX 9070 GRE - this spells good news for folks on platforms that might still use older generations of PCIe interfaces (e.g. PCIe 4.0 x16 and even PCIe 3.0 x16), where the performance impact of using the RX 9070 GRE on such older platforms would likely be smaller than say, using a PCIe 5.0 x8 RTX 5060 Ti graphics card on a PC that supports PCIe 3.0 (which would mean that the said RTX 5060 Ti would operate at PCIe 3.0 x8 mode instead - the 9070 GRE would run at PCIe 3.0 x16 given that it has the full electrical pin-outs for 16 lanes).

Display outputs wise, the Swift Radeon RX 9070 GRE comes with three DisplayPort 2.1a ports and a single HDMI 2.1b port.

GPU-Z Screenshot

Benchmarks & Test Setup

To get a sensing of the card's gaming performance, the following in-game benchmark tests were run:

• Assassin's Creed Mirage in-game benchmark
• Black Myth Wukong Benchmark Tool
• Cyberpunk 2077 v2.31 in-game benchmark
• F1 23 in-game benchmark
• Far Cry 6 in-game benchmark
• Shadow of the Tomb Raider in-game benchmark

The benchmark tests above were run at both the 1440p and 1080p resolution. Upscaling technologies such as FSR as well as Frame Generation were also disabled for the test runs.

Test Setup

A test bench with the specifications as listed below was used:

• CPU: AMD Ryzen 7 7800X3D (stock settings, PBO & Curve Optimiser disabled)
• CPU Cooler: Lian Li HydroShift II LCD-S 360 TL
• CPU Thermal Paste: Arctic MX-4
• Motherboard: Gigabyte X670E AORUS MASTER (F42b BIOS)
• Memory: 32GB (2x16GB) KingBank DDR5 Dark Heatsink UDIMM 6000MHz (DDR5 6000 CL28-35-35-76 1.45V)
• Storage: Lexar NM790 1TB NVMe SSD
• Graphics XFX Swift Radeon RX 9070 GRE Triple Fan Gaming Edition 12GB GDDR6
• PSU: Lian Li SX1200P
• Case: Lian Li LANCOOL 217 with 3x Lian Li Uni Fan SL-Inf Wireless 120 Reverse blade, 1x Lian Li Uni Fan SL-Inf Wireless 140 and JEYI ARGB PWM 5-way fan hub
• Monitor: Gigabyte M27U 27" LCD monitor (4K, 160Hz)
• OS: Microsoft Windows 11 Pro (25H2 with all latest drivers and updates installed, "Balanced" power profile)
• Graphics Driver: AMD Adrenalin Edition 26.5.2
• Resizable BAR: Enabled

The test results shown below were taken from benchmark runs that did not have any screen capture software running in the background.

Assassin's Creed Mirage In-game Benchmark

First up in the benchmarks section is Assassin's Creed Mirage's in-game benchmark test, where on the 1440p resolution, the test bench achieves an average FPS of 115, with a minimum FPS of 38 and a maximum FPS of 257. These results were obtained on the "Ultra High" preset with Adaptive Quality set to "60 FPS", and all forms of upscaling (e.g. FSR) disabled.

On the 1080p resolution with the same quality settings as the 1440p run, the test bench puts out an average FPS of 144, a minimum FPS of 46 and a maximum FPS of 279.

Black Myth Wukong Benchmark Tool

Next up is Black Myth Wukong's Benchmark Tool next, where at 1440p, the test bench achieves an average FPS of 62, with a minimum FPS of 51 and a maximum FPS of 71. These results were obtained on the "High" preset with ray tracing switched off, and Super Resolution set to "100" with TSR as the sampling mode (i.e. FSR was disabled). Frame Generation was also disabled for the duration of the benchmark.

Bumping the resolution down to 1080p with the same graphics quality settings as the 1440p benchmark run, the test system achieves an average FPS of 89, a minimum FPS of 63 and a maximum FPS of 105.

Cyberpunk 2077 v2.31 In-game Benchmark

Moving on to Cyberpunk 2077 v2.31's in-game benchmark test, the test bench achieves an average FPS of 85.78, a minimum FPS of 72.37 and a maximum FPS of 105.44 on the 1440p resolution. These results were obtained on the "Ray Tracing Low" preset with FSR and Frame Generation disabled.

On the 1080p resolution with the same quality settings as the 1440p test run, the test bench puts out an average FPS of 127.52, a minimum FPS of 108.58 and a maximum FPS of 166.32.

F1 23 In-game Benchmark

With F1 23's in-game benchmark tool, the test system achieves an average FPS of 67, a minimum FPS of 52 and a maximum FPS of 87 at 1440p. These results were obtained on the "Ultra High" detail preset with all forms of upscaling technologies (e.g. FSR) disabled. Anisotropic Filtering was set to the "16x" setting.

Turning down the resolution to 1080p while maintaining the same graphics quality settings, the test system achieves an average FPS of 101, a minimum FPS of 81 and a maximum FPS of 129.

Far Cry 6 In-game Benchmark

Far Cry 6's in-game benchmark sees the test bench achieves an average FPS of 136, a minimum FPS of 118 and a maximum FPS of 159 at 1440p. These results were obtained on the "Ultra" quality preset with all forms of upscaling disabled. and DXR reflections and shadows were enabled throughout the duration of testing.

Bumping the resolution down to 1080p with the same quality settings as the 1440p benchmark run, the test bench achieves an average FPS of 159, a minimum FPS of 127 and a maximum FPS of 205.

Shadow of the Tomb Raider In-game Benchmark

Rounding things off in the benchmarks section is Shadow of the Tomb Raider's built-in benchmark test, where at 1440p, the test bench achieves an average FPS of 78. The results above were obtained on maxed out graphics settings, where Ray Tracing Shadow Quality was set to "Ultra" and all forms of upscaling technologies such as FSR were disabled.

On the 1080p resolution and retaining the same graphics settings as the 1440p test run, the test system achieves an average FPS of 117.

Power Consumption, Thermals & Acoustics

Power Consumption

Under Load (F1 23 In-game Benchmark, 1440p, Ultra High Detail Preset, TAA, 16x AF, FSR Off, Singapore Circuit, Five Laps)
• Avg. Total Board Power: 218.9W
• Min. Total Board Power: 183.5W
• Max. Total Board Power: 219.6W

Idle
• Avg. Total Board Power: 5.5W
• Min. Total Board Power: 4.8W
• Max. Total Board Power: 18.1W

The figures above were obtained from "Total Board Power" readouts from HWiNFO64.

Thermals

Under Load (F1 23 In-game Benchmark, 1440p, Ultra High Detail Preset, TAA, 16x AF, FSR Off, Singapore Circuit, Five Laps)

• Avg. GPU Temperature: 62.1°C
• Min. GPU Temperature: 54.0°C
• Max. GPU Temperature: 64.0°C
• Avg. GPU Memory Junction Temperature: 71.5°C
• Min. GPU Memory Junction Temperature: 62.0°C
• Max. GPU Memory Junction Temperature: 74.0°C

Idle
• Avg. GPU Temperature: 36.3°C
• Min. GPU Temperature: 36.0°C
• Max. GPU Temperature: 37.0°C
• Avg. GPU Memory Junction Temperature: 36.0°C
• Min. GPU Memory Junction Temperature: 36.0°C
• Max. GPU Memory Junction Temperature: 38.0°C

The figures above were obtained from "GPU Temperature" & "GPU Memory Junction Temperature" readouts from HWiNFO64. Do note that the benchmarks were carried out in a non-airconditioned room with an ambient temperature of around 30°C.

While under gaming workloads with default fan curve settings, I found the XFX Swift Radeon RX 9070 GRE Triple Fan 12GB GDDR6's fans to be reasonably quiet. That said, there was some amount of coil whine, where the said coil whine was not very loud and manifested itself in the form of a low buzzing noise that could be heard with the side panel of the PC case removed - on a scale of zero to five, with zero being no coil whine at all, I'd rate the amount of coil whine with this card a 2/5.

Conclusion & Pricing

RTX 5060 Ti 16GB vs RX9070 GRE - Six-Game Comparison at 1440p & 1080p

To better assess the value of the RX 9070 GRE in the context of the GPU market, I've also put NVIDIA's GeForce RTX 5060 Ti 16GB through the same set of benchmarks, and the results can be seen above (for added information, the RTX 5060 Ti 16GB results were obtained by testing a MSI GeForce RTX 5060 Ti 16G GAMING TRIO OC on the same test bench PC). The long and short of it is, the 9070 GRE is around 10% to 20% faster than the RTX 5060 Ti 16GB, at least from the small sample size of the six gaming benchmark tests that I ran at 1440p and 1080p.

In AMD’s reviewer’s guide and marketing materials for the 9070 GRE, the GPU is pitted against the RTX 5060 Ti 16GB, reason being that although the 5060 Ti 16GB first launched with a MSRP of US$429 back in April 2025, the whole AI data centre boom has since sent RAM prices skywards, and graphics cards with GDDR7 memory modules have seen their prices go up as well, the 5060 Ti 16GB included; as of end-May 2026, the cheapest 5060 Ti 16GB graphics cards are going for around US$569 or higher on e-commerce platforms such as Amazon US and Newegg US.

AMD touts the RX 9070 GRE to have a launch MSRP of US$549, and assuming that the 9070 GRE does indeed stay at MSRP levels post-launch, getting it over the 5060 Ti 16GB does seem like a no-brainer, especially given that it does perform better than the 5060 Ti 16GB after all (once again, from the six games that I’ve tested, small size of benchmarks I know). With this in mind, one can almost see why AMD decided to launch the 9070 GRE for the global market, seeing how uncompetitively priced the RTX 5060 Ti 16GB is as of the time of writing this article.

For added context, the RX 9060 XT 16GB, which is a tier down from the RX 9070 GRE, has the cheapest models going for around US$439 (as of end-May 2026) on platforms such as Newegg US, and this said GPU performs very similarly to the 5060 Ti 16GB too from a gaming perspective. On the other hand, the cheapest 9070 (non-GRE non-XT) graphics card is going for US$599 as of end-May 2026.

All in all, the 9070 GRE does slot in nicely between AMD’s 9060 XT 16GB and the 9070 non-XT GPUs, and given current market conditions for DIY PC builders, the 9070 GRE does at the bare minimum, give consumers an additional GPU option to choose from to better meet their needs, where it's a capable GPU that allows you to game comfortably at 1440p and 1080p resolutions.

Get this graphics card from Amazon: https://amzn.to/4dJJO0I (affiliate link)

ALKtech may earn commissions from qualifying purchases made using affiliate links, at no extra cost to you as a customer. These commissions go a long way in supporting the platform!

The graphics card featured in this article was a review unit provided on loan from AMD Singapore.