Conclusion - high performance with avoidable limitations
With the AMD Ryzen AI 9 HX 470, the Geekom A9 Max is currently one of the more powerful mini PCs in a compact format and offers a solid basis for productive applications and AI-supported tasks. The equipment includes 32 GB DDR5 RAM and a 2 TB SSD, whereby the selected RAM configuration - similar to the recently tested Minisforum AI X1 Pro - ensures a somewhat unbalanced overall performance. The memory connection in particular limits the system in places and prevents the CPU and iGPU from reaching their full potential.
Nevertheless, the Geekom A9 Max is well positioned in terms of connectivity and offers numerous modern interfaces, such as two USB4 interfaces. However, users still have to do without features such as OCuLink or USB4 v2. The workmanship and material quality leave a high-quality impression and fit in well with the premium claim of the Geekom A9 Max. However, the casing design is already familiar from other models from the manufacturer, which is why a redesign would definitely make sense - especially to integrate more modern USB-C ports at the front.
Another positive aspect is the overall good energy efficiency. The mini PC is economical in everyday use and remains comparatively quiet under load. Although the cooling system is slightly audible when idling, the background noise is not disturbing overall. As a compact gaming PC, the Geekom A9 Max can only convince to a limited extent in the tested configuration. Only with a second RAM module does the 3D performance develop in the right direction, even if systems with a dedicated graphics card still have a clear advantage.
Pros
Cons
Price and availability
The Geekom A9 Max is available directly from the manufacturer and via Amazon at a price of $1,799. With the discount code "NKA9268F", the price is reduced by an additional 8 percent. The voucher is valid until November 1st and can be redeemed both in the Geekom online shop and on Amazon.
With the A9 Max in the 2026 edition, Geekom is expanding its range with another powerful mini PC for demanding users. Inside is the AMD Ryzen AI 9 HX 470 from the current Gorgon Point series, which promises high computing power with 12 cores and 24 threads. The APU is complemented by the integrated Radeon 890M graphics unit based on RDNA 3.5 and the XDNA2 NPU for AI-supported applications and modern AI features.
Specifications
Case & Connectivity
The Geekom A9 Max is based on the same housing concept as the Geekom A8 Max and the variant with AMD Ryzen 9 HX 370. Further details can be found in the respective tests.
USB4-Speed (ADATA SE920)
| Connectivity - USB 3.x / USB4 / TB - AS SSD Seq Read Test (1GB) | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Geekom A9 Max, AI 9 HX 370 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Khadas Mind Pro | |
| Average of class Mini PC (423 - 3906, n=43, last 2 years) | |
SD card reader
| SD Card Reader | |
| average JPG Copy Test (av. of 3 runs) | |
| Khadas Mind (AV Pro SD microSD 128 GB V60) | |
| Geekom A9 Max, AI 9 HX 470 (AV Pro SD microSD 128 GB V60) | |
| Geekom A5 (AV Pro SD microSD 128 GB V60) | |
| Geekom Mini IT13 | |
| Geekom Mini IT12 (AV Pro SD microSD 128 GB V60) | |
| Average of class Mini PC (15.1 - 214, n=25, last 2 years) | |
| Zotac ZBOX Magnus EN173070C (AV Pro SD microSD 128 GB V60) | |
| maximum AS SSD Seq Read Test (1GB) | |
| Khadas Mind (AV Pro SD microSD 128 GB V60) | |
| Geekom A9 Max, AI 9 HX 470 (AV Pro SD microSD 128 GB V60) | |
| Geekom A5 (AV Pro SD microSD 128 GB V60) | |
| Geekom Mini IT13 | |
| Geekom Mini IT12 (AV Pro SD microSD 128 GB V60) | |
| Average of class Mini PC (17.9 - 250, n=25, last 2 years) | |
| Zotac ZBOX Magnus EN173070C (AV Pro SD microSD 128 GB V60) | |
Wi-Fi performance
Accessories
Maintenance
The Mini PC arrives to the user already equipped with RAM and SSD, so no additional hardware setup is necessary. Components such as RAM, mass storage and WLAN module can be replaced if necessary, even if access to the housing is not particularly convenient.
Sustainability
The Geekom A9 Max offers only limited options for repairs or extensive maintenance work, but comes with a 36-month manufacturer's warranty. The space-saving packaging also makes the device ideal for transportation. In terms of certifications, Geekom mentions CE, FCC, CB, CCC, SRRC, and RoHS, among others, but does not provide further details on sustainable or recycled materials.
Test conditions
All measurements were carried out with the default settings and activated normal mode. A performance and silent mode are also available in the BIOS, with the latter reducing the CPU's power consumption slightly.
Processor
The processor installed in the Geekom A9 Max AMD Ryzen AI 9 HX 470 offers high computing power overall but cannot clearly distinguish itself from the AMD Ryzen AI 9 HX 370 in the test. In addition to possible fluctuations between individual processors, the memory configuration in particular is likely to influence the results, as can be seen several times in the course of the benchmarks. Equipping the system with only one RAM module noticeably limits the system and already costs some performance in the CPU tests. In terms of power consumption, the Ryzen AI 9 HX 470 is on a similar level to comparable systems with Ryzen AI 9 HX 370, with up to 65 watts for short periods and around 54 watts under continuous load.
| Performance rating - Percent | |
| GMK EVO-T1 | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A8 Max, R7 8845HS | |
| Cinebench R10 | |
| Rendering Single CPUs 64Bit | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Rendering Multiple CPUs 64Bit | |
| GMK EVO-T1 | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| WinRAR - Result | |
| Geekom A8 Max, R7 8845HS | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| 3DMark 11 - 1280x720 Performance Physics | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| HWBOT x265 Benchmark v2.2 - 4k Preset | |
| Geekom A9 Max, AI 9 HX 370 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| GMK EVO-T1 | |
| Khadas Mind Pro | |
| Geekom A8 Max, R7 8845HS | |
| R Benchmark 2.5 - Overall mean | |
| Geekom A8 Max, R7 8845HS | |
| GMK EVO-T1 | |
| Khadas Mind Pro | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| LibreOffice - 20 Documents To PDF | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Khadas Mind Pro | |
| Super Pi mod 1.5 XS 1M - 1M | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| GMK EVO-T1 | |
| Khadas Mind Pro | |
| Geekom A8 Max, R7 8845HS | |
| Super Pi mod 1.5 XS 2M - 2M | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| GMK EVO-T1 | |
| Khadas Mind Pro | |
| Affinity Photo 2 - Combined (Multi CPU) | |
| GMK EVO-T1 | |
| Khadas Mind Pro | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
* ... smaller is better
AIDA64: FP32 Ray-Trace | FPU Julia | CPU SHA3 | CPU Queen | FPU SinJulia | FPU Mandel | CPU AES | CPU ZLib | FP64 Ray-Trace | CPU PhotoWorxx | CPU CheckMate
| Performance rating | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A8 Max, R7 8845HS -1! | |
| Average of class Mini PC | |
| Khadas Mind Pro -1! | |
| GMK EVO-T1 | |
| AIDA64 / FP32 Ray-Trace | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| Average of class Mini PC (974 - 82725, n=108, last 2 years) | |
| GMK EVO-T1 (v6.92.6600) | |
| GMK EVO-T1 (v8.20.8100) | |
| Khadas Mind Pro (v8.20.8100) | |
| AIDA64 / FPU Julia | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| GMK EVO-T1 (v6.92.6600) | |
| GMK EVO-T1 (v8.20.8100) | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| Average of class Mini PC (6115 - 363782, n=108, last 2 years) | |
| Khadas Mind Pro (v8.20.8100) | |
| AIDA64 / CPU SHA3 | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| GMK EVO-T1 (v6.92.6600) | |
| GMK EVO-T1 (v8.20.8100) | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| Khadas Mind Pro (v8.20.8100) | |
| Average of class Mini PC (385 - 9475, n=108, last 2 years) | |
| AIDA64 / CPU Queen | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| Average of class Mini PC (20501 - 184747, n=94, last 2 years) | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| GMK EVO-T1 (v6.92.6600) | |
| AIDA64 / FPU SinJulia | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| Average of class Mini PC (1019 - 31931, n=108, last 2 years) | |
| GMK EVO-T1 (v6.92.6600) | |
| GMK EVO-T1 (v8.20.8100) | |
| Khadas Mind Pro (v8.20.8100) | |
| AIDA64 / FPU Mandel | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| GMK EVO-T1 (v6.92.6600) | |
| GMK EVO-T1 (v8.20.8100) | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| Average of class Mini PC (3322 - 192994, n=108, last 2 years) | |
| Khadas Mind Pro (v8.20.8100) | |
| AIDA64 / CPU AES | |
| Average of class Mini PC (9214 - 325960, n=107, last 2 years) | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| GMK EVO-T1 (v6.92.6600) | |
| GMK EVO-T1 (v8.20.8100) | |
| Khadas Mind Pro (v8.20.8100) | |
| AIDA64 / CPU ZLib | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Khadas Mind Pro (v8.20.8100) | |
| GMK EVO-T1 (v6.92.6600) | |
| GMK EVO-T1 (v8.20.8100) | |
| Average of class Mini PC (127.2 - 2431, n=108, last 2 years) | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| AIDA64 / FP64 Ray-Trace | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| Average of class Mini PC (537 - 43587, n=108, last 2 years) | |
| GMK EVO-T1 (v6.92.6600) | |
| GMK EVO-T1 (v8.20.8100) | |
| Khadas Mind Pro (v8.20.8100) | |
| AIDA64 / CPU PhotoWorxx | |
| Khadas Mind Pro (v8.20.8100) | |
| GMK EVO-T1 (v8.20.8100) | |
| GMK EVO-T1 (v6.92.6600) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v6.92.6600) | |
| Geekom A8 Max, R7 8845HS (v6.92.6600) | |
| Average of class Mini PC (8069 - 96843, n=108, last 2 years) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v6.92.6600) | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Minisforum AI X1 Pro, HX 470 (v6.92.6600) | |
| AIDA64 / CPU CheckMate | |
| Minisforum AI X1 Pro, HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 470 (v8.20.8100) | |
| Geekom A9 Max, AI 9 HX 370 (v8.20.8100) | |
| Average of class Mini PC (1743 - 19453, n=14, last 2 years) | |
| Khadas Mind Pro (v8.20.8100) | |
| GMK EVO-T1 (v8.20.8100) | |
System performance
With its powerful APU, the Geekom A9 Max provides a solid foundation but cannot fully utilize the available hardware in its factory settings. The memory connection in particular limits the performance somewhat, which is particularly noticeable in the RAM benchmarks. Comparable results were also shown by the Minisforum AI X1 Pro, which was also only equipped with a single RAM module in our test. As a result, the overall performance in some disciplines falls slightly short of what is technically possible.
WebXPRT 4: Overall
WebXPRT 3: Overall
Mozilla Kraken 1.1: Total
CrossMark: Overall | Productivity | Creativity | Responsiveness
* ... smaller is better
* ... smaller is better
| UL Procyon for Windows / Overall Score Integer NPU | |
| Khadas Mind Pro (OpenVINO) | |
| Minisforum AI X1 Pro, HX 470 (Ryzen AI) | |
| Geekom A9 Max, AI 9 HX 470 (Ryzen AI) | |
| Geekom A9 Max, AI 9 HX 370 (AMD Ryzen AI) | |
| GMK EVO-T1 (OpenVINO) | |
| UL Procyon for Windows / Overall Score Integer GPU | |
| Khadas Mind Pro (Windows ML) | |
| GMK EVO-T1 (Windows ML) | |
| Geekom A9 Max, AI 9 HX 370 (Windows ML) | |
| Minisforum AI X1 Pro, HX 470 (Windows ML) | |
| Geekom A9 Max, AI 9 HX 470 (Windows ML) | |
DPC latencies
| DPC Latencies / LatencyMon - interrupt to process latency (max), Web, Youtube, Prime95 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| GMK EVO-T1 | |
| Khadas Mind Pro | |
* ... smaller is better
Storage devices
| Drive Performance rating - Percent | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Khadas Mind Pro | |
| Average of class Mini PC | |
| Geekom A8 Max, R7 8845HS | |
| Minisforum AI X1 Pro, HX 470 | |
* ... smaller is better
Disk Throttling: DiskSpd Read Loop, Queue Depth 8
Graphics card and gaming performance
The Geekom A9 Max features the AMD Radeon 890M is used as the integrated graphics unit based on the RDNA3+ architecture. Compared to the Radeon 780M, it basically offers more performance, but cannot fully develop its potential in the test. The reason for this is - as with the recently tested Minisforum AI X1 Pro - the single-channel RAM configuration of the test system. As a result, the iGPU falls short of expectations in some benchmarks and is sometimes even outperformed by older solutions. With an upgrade to dual-channel RAM, however, the graphics performance can be noticeably increased, as can be seen in this accompanying article can read.
| 3DMark Performance rating - Percent | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| 3DMark 11 - 1280x720 Performance GPU | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| Unigine Valley 1.0 | |
| 1920x1080 Extreme HD DirectX AA:x8 | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| 1920x1080 Extreme HD Preset OpenGL AA:x8 | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Affinity Photo 2 - Combined (Single / Multi GPU) | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A8 Max, R7 8845HS | |
SPECworkstation 4.0: Graphics Score | Accelerator Score | Viewport Graphics | HandBrake | ONNX Inference
| Performance rating | |
| Average NVIDIA GeForce RTX 4050 Laptop GPU | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| SPECworkstation 4.0 / Graphics Score | |
| Average NVIDIA GeForce RTX 4050 Laptop GPU (n=1) | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Minisforum AI X1 Pro, HX 470 | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 470 | |
| GMK EVO-T1 | |
| SPECworkstation 4.0 / Accelerator Score | |
| Average NVIDIA GeForce RTX 4050 Laptop GPU (n=1) | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| SPECworkstation 4.0 / Viewport Graphics | |
| Average NVIDIA GeForce RTX 4050 Laptop GPU (n=1) | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Minisforum AI X1 Pro, HX 470 | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 470 | |
| GMK EVO-T1 | |
| SPECworkstation 4.0 / HandBrake | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Minisforum AI X1 Pro, HX 470 | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A8 Max, R7 8845HS | |
| Average NVIDIA GeForce RTX 4050 Laptop GPU (n=1) | |
| SPECworkstation 4.0 / ONNX Inference | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Average NVIDIA GeForce RTX 4050 Laptop GPU (n=1) | |
| Geekom A9 Max, AI 9 HX 370 | |
| Khadas Mind Pro | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A9 Max, AI 9 HX 470, AMD Radeon 890M Ryzen AI 9 HX 470 | Khadas Mind Pro, Intel Arc B390 12 Xe3 Panther Lake iGPU Ultra X7 358H | Geekom A8 Max, R7 8845HS, AMD Radeon 780M R7 8845HS | Geekom A9 Max, AI 9 HX 370, AMD Radeon 890M Ryzen AI 9 HX 370 | Minisforum AI X1 Pro, HX 470, AMD Radeon 890M Ryzen AI 9 HX 470 | GMK EVO-T1, Intel Arc Graphics 140T Ultra 9 285H | Average of class Mini PC | |
|---|---|---|---|---|---|---|---|
| SPECviewperf 2020 | 2% | 48% | 62% | 3% | -29% | 130% | |
| 3840x2160 Medical (medical-03) (fps) | 24.6 | 9.48 -61% | 12.7 -48% | 26.1 6% | 23.9 -3% | 7.02 -71% | 22.2 ? -10% |
| 3840x2160 Energy (energy-03) (fps) | 14.5 | 7.67 -47% | 21.2 46% | 18 24% | 14 -3% | 7.4 -49% | 32.8 ? 126% |
| 3840x2160 Creo (creo-03) (fps) | 18.5 | 28.9 56% | 34.9 89% | 36.2 96% | 21.4 16% | 19.9 8% | 49.1 ? 165% |
| 3840x2160 CATIA (catia-06) (fps) | 16.1 | 19.5 21% | 25 55% | 26 61% | 16.2 1% | 12.8 -20% | 34.2 ? 112% |
| 3840x2160 Siemens NX (snx-04) (fps) | 35.8 | 7.09 -80% | 66.2 85% | 70.6 97% | 39.2 9% | 6.13 -83% | 52.7 ? 47% |
| 3840x2160 Maya (maya-06) (fps) | 31.2 | 69.4 122% | 50.2 61% | 59.3 90% | 31.3 0% | 44.9 44% | 136.7 ? 338% |
| 3DMark 05 Standard | 47050 points | |
| 3DMark 06 Standard Score | 32553 points | |
| 3DMark Vantage P Result | 27683 points | |
| 3DMark 11 Performance | 9201 points | |
| 3DMark Ice Storm Standard Score | 140116 points | |
| 3DMark Cloud Gate Standard Score | 26859 points | |
| 3DMark Fire Strike Score | 4925 points | |
| 3DMark Fire Strike Extreme Score | 2716 points | |
| 3DMark Time Spy Score | 2168 points | |
| 3DMark Steel Nomad Score | 369 points | |
Help | ||
| GTA V | |
| 1024x768 Lowest Settings possible | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| 1920x1080 High/On (Advanced Graphics Off) AA:2xMSAA + FX AF:8x | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| Cyberpunk 2077 | |
| 1920x1080 Low Preset (FSR off) | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| 1920x1080 High Preset (FSR off) | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| GMK EVO-T1 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Baldur's Gate 3 | |
| 1280x720 Low Preset AA:SM | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| 1920x1080 High Preset AA:T | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| F1 24 | |
| 1920x1080 Low Preset AA:T AF:16x | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| 1920x1080 High Preset AA:T AF:16x | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
FPS chart - 30 min Cyberpunk 2077 1920 x 1080 Ultra Preset (FSR off)
| low | med. | high | ultra | |
|---|---|---|---|---|
| GTA V (2015) | 135 | 125.9 | 39.1 | 16.2 |
| The Witcher 3 (2015) | 97.2 | 61.4 | 31.3 | 21.5 |
| Dota 2 Reborn (2015) | 160.6 | 111.8 | 61.8 | 55.4 |
| Final Fantasy XV Benchmark (2018) | 40.9 | 24.6 | 18.4 | |
| Baldur's Gate 3 (2023) | 27.9 | 21.5 | 18.5 | 18.3 |
| Cyberpunk 2077 (2023) | 29.4 | 16.9 | 13 | 16.5 |
| F1 24 (2024) | 53.7 | 47.9 | 31.1 | 10.2 |
Emissions
System noise
For the cooling system, Geekom continues to rely on the familiar concept that is already used in a similar form in other models in the series. In idle mode, the Geekom A9 Max is slightly louder than the identical device with AMD Ryzen AI 9 HX 370, although the system is much more restrained under sustained load. Even under high load, the overall noise development remains at a pleasant level. The fan control reacts in a predominantly balanced manner, but could be even more finely tuned in some situations, especially under medium load.
Noise level
| Idle |
| 27.3 / 27.4 / 27.5 dB(A) |
| Load |
| 43.4 / 43.2 dB(A) |
![]() | ||
30 dB silent 40 dB(A) audible 50 dB(A) loud |
||
min: | ||
| Geekom A9 Max, AI 9 HX 470 Ryzen AI 9 HX 470, Radeon 890M | Khadas Mind Pro Ultra X7 358H, Arc B390 Panther Lake iGPU | Geekom A8 Max, R7 8845HS R7 8845HS, Radeon 780M | Geekom A9 Max, AI 9 HX 370 Ryzen AI 9 HX 370, Radeon 890M | Minisforum AI X1 Pro, HX 470 Ryzen AI 9 HX 470, Radeon 890M | GMK EVO-T1 Ultra 9 285H, Arc 140T | |
|---|---|---|---|---|---|---|
| Noise | 9% | 7% | -5% | 9% | -21% | |
| off / environment * (dB) | 24.7 | 24.6 -0% | 23.8 4% | 24.8 -0% | 24.9 -1% | 24.7 -0% |
| Idle Minimum * (dB) | 27.3 | 24.9 9% | 24 12% | 25.4 7% | 28.4 -4% | 38.7 -42% |
| Idle Average * (dB) | 27.4 | 24.9 9% | 24.1 12% | 25.4 7% | 28.6 -4% | 38.8 -42% |
| Idle Maximum * (dB) | 27.5 | 25 9% | 24.1 12% | 25.5 7% | 28.6 -4% | 38.9 -41% |
| Load Average * (dB) | 43.4 | 40.1 8% | 42.7 2% | 49.8 -15% | 33.1 24% | 51.6 -19% |
| Witcher 3 ultra * (dB) | 43.4 | 36.2 17% | 49.7 -15% | 32.3 26% | 38.7 11% | |
| Cyberpunk 2077 ultra * (dB) | 43.6 | 39.7 9% | 42.5 3% | 49.6 -14% | 32.7 25% | 52 -19% |
| Load Maximum * (dB) | 43.2 | 40.2 7% | 42.6 1% | 49.8 -15% | 39.2 9% | 51.6 -19% |
* ... smaller is better
Temperature
The casing temperatures of the Geekom A9 Max remain inconspicuous overall in everyday use and under load. Even during longer load phases, the casing only heats up moderately. In the stress test, the processor briefly reaches higher temperatures of up to 90 °C, which is harmless. This then stabilizes at around 85 °C over the course of the test. The cooling system cannot maintain the maximum power consumption permanently, which is why the CPU stabilizes at a slightly lower continuous power after a while.
(+) The maximum temperature on the upper side is 33.4 °C / 92 F, compared to the average of 35.5 °C / 96 F, ranging from 20.6 to 69.5 °C for the class Mini PC.
(+) The bottom heats up to a maximum of 36.9 °C / 98 F, compared to the average of 36.2 °C / 97 F
(+) In idle usage, the average temperature for the upper side is 26.4 °C / 80 F, compared to the device average of 32.2 °C / 90 F.
(+) 3: The average temperature for the upper side is 31.2 °C / 88 F, compared to the average of 32.2 °C / 90 F for the class Mini PC.
Energy management
With the AMD Ryzen AI 9 HX 470, the Geekom A9 Max offers high computing power with moderate energy requirements overall. In the efficiency comparison, the system shows solid values but without clearly standing out from similar models with a Ryzen AI APU.
Overall, the power consumption remains pleasantly low for a mini PC. The Geekom A9 Max only consumes a few watts in idle mode, while it reaches around 73 watts under load. Short-term load peaks are slightly higher but remain within the expected range for the compact system.
| Off / Standby | |
| Idle | |
| Load |
|
Key:
min: | |
| Geekom A9 Max, AI 9 HX 470 Ryzen AI 9 HX 470, Radeon 890M | Khadas Mind Pro Ultra X7 358H, Arc B390 Panther Lake iGPU | Geekom A8 Max, R7 8845HS R7 8845HS, Radeon 780M | Geekom A9 Max, AI 9 HX 370 Ryzen AI 9 HX 370, Radeon 890M | Minisforum AI X1 Pro, HX 470 Ryzen AI 9 HX 470, Radeon 890M | GMK EVO-T1 Ultra 9 285H, Arc 140T | Average AMD Radeon 890M | Average of class Mini PC | |
|---|---|---|---|---|---|---|---|---|
| Power Consumption | 5% | -14% | -11% | -26% | -49% | -33% | -87% | |
| Idle Minimum * (Watt) | 3.7 | 3.9 -5% | 3.1 16% | 2.4 35% | 5.5 -49% | 6.1 -65% | 5.63 ? -52% | 8.73 ? -136% |
| Idle Average * (Watt) | 4.5 | 4.7 -4% | 5.7 -27% | 4.6 -2% | 7.7 -71% | 7.9 -76% | 7.81 ? -74% | 10.9 ? -142% |
| Idle Maximum * (Watt) | 5.7 | 5.2 9% | 8.3 -46% | 9.2 -61% | 9.1 -60% | 13.2 -132% | 9.33 ? -64% | 14 ? -146% |
| Load Average * (Watt) | 72.6 | 42.6 41% | 68.3 6% | 78.4 -8% | 59.2 18% | 60.7 16% | 63.7 ? 12% | 64.1 ? 12% |
| Cyberpunk 2077 ultra external monitor * (Watt) | 62.8 | 67.5 -7% | 78.7 -25% | 77.8 -24% | 57.4 9% | 67.4 -7% | ||
| Load Maximum * (Watt) | 90.4 | 93.5 -3% | 97.7 -8% | 95.1 -5% | 92.4 -2% | 117.1 -30% | 78.7 ? 13% | 112 ? -24% |
* ... smaller is better
Power efficiency
| Performance rating - Percent | |
| Khadas Mind Pro | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Power Consumption | |
| Cinebench R23 Multi Power Efficiency - external Monitor | |
| Geekom A9 Max, AI 9 HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Minisforum AI X1 Pro, HX 470 | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Cinebench R23 Single Power Efficiency - external Monitor | |
| Khadas Mind Pro | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A9 Max, AI 9 HX 470 | |
| GMK EVO-T1 | |
| Geekom A8 Max, R7 8845HS | |
| Cinebench R15 Multi Efficiency (external Monitor) | |
| Geekom A9 Max, AI 9 HX 470 | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Khadas Mind Pro | |
| Geekom A8 Max, R7 8845HS | |
| GMK EVO-T1 | |
| Cyberpunk 2077 ultra Efficiency | |
| Khadas Mind Pro | |
| GMK EVO-T1 | |
| Geekom A9 Max, AI 9 HX 370 | |
| Geekom A8 Max, R7 8845HS | |
| Minisforum AI X1 Pro, HX 470 | |
| Geekom A9 Max, AI 9 HX 470 | |
energy consumption with external monitor
energy consumption with external monitor
Notebookcheck rating
Geekom A9 Max, AI 9 HX 470
- 05/23/2026 v8
Sebastian Bade
Possible alternatives in comparison
Image | Model / Review | Price | Weight | Dimensions |
|---|---|---|---|---|
| Geekom A9 Max, AI 9 HX 470 AMD Ryzen AI 9 HX 470 ⎘ AMD Radeon 890M ⎘ 32 GB Memory, 2048 GB SSD | Amazon: List Price: 1699€ | 686 g | 135 x 135 x 46.9 mm | |
| Khadas Mind Pro Intel Core Ultra X7 358H ⎘ Intel Arc B390 12 Xe3 Panther Lake iGPU ⎘ 64 GB Memory, 2048 GB SSD | Amazon: 1. $1,099.00 Khadas Mind 2 Mini PC - Inte... 2. $1,399.00 Khadas Mind 2 Mini PC - Inte... 3. $549.00 Khadas Mind 1 Mini PC, Intel... List Price: 1799€ | 566 g | 105 x 146 x 58 mm | |
| Geekom A8 Max, R7 8845HS AMD Ryzen 7 8845HS ⎘ AMD Radeon 780M ⎘ 32 GB Memory, 1024 GB SSD | Amazon: List Price: 899 EUR | 681 g | 132 x 135 x 46.9 mm | |
| Geekom A9 Max, AI 9 HX 370 AMD Ryzen AI 9 HX 370 ⎘ AMD Radeon 890M ⎘ 32 GB Memory, 2048 GB SSD | Amazon: List Price: 999€ | 709 g | 132 x 135 x 46.9 mm | |
| Minisforum AI X1 Pro, HX 470 AMD Ryzen AI 9 HX 470 ⎘ AMD Radeon 890M ⎘ 32 GB Memory, 1024 GB SSD | Amazon: List Price: 1399€ | 1.3 kg | 195 x 195 x 42.5 mm | |
| GMK EVO-T1 Intel Core Ultra 9 285H ⎘ Intel Arc Graphics 140T ⎘ 64 GB Memory, 1024 GB SSD | Amazon: List Price: 999€ | 946 g | 151 x 154 x 73.6 mm |
Transparency
The selection of devices to be reviewed is made by our editorial team. The test sample was provided to the author as a loan by the manufacturer or retailer for the purpose of this review. The lender had no influence on this review, nor did the manufacturer receive a copy of this review before publication. There was no obligation to publish this review. As an independent media company, Notebookcheck is not subjected to the authority of manufacturers, retailers or publishers.
This is how Notebookcheck is testing
Every year, Notebookcheck independently reviews hundreds of laptops and smartphones using standardized procedures to ensure that all results are comparable. We have continuously developed our test methods for around 20 years and set industry standards in the process. In our test labs, high-quality measuring equipment is utilized by experienced technicians and editors. These tests involve a multi-stage validation process. Our complex rating system is based on hundreds of well-founded measurements and benchmarks, which maintains objectivity. Further information on our test methods can be found here.
































































