StorageReview measured 7,306 points for the ASUS ProArt GR1X at the end of a 30-minute Cinebench 2026 loop. In a separate local-AI test, the same 64GB/1TB system produced 40.0 tokens per second serving Qwen3.8 27B with eight concurrent requests. The clearest differences depend on the workload: the CPU loop and concurrent AI results favor the GR1X over the tested laptops, while one-request results are much closer. This follow-up adds detailed benchmark results to NeoTeo’s earlier ProArt GR1X coverage.

Which GR1X configuration produced these results?

The benchmarked GR1X had 64GB of unified memory, a 1TB Micron PCIe Gen4 SSD and Windows 11 Pro 26H1 on Arm. StorageReview ran it in Best performance mode with a 4K external monitor attached.

The laptop comparisons used sr-client 1.11, while the GR1X and MSI desktop tests used version 1.13. The systems also differed in memory and storage configurations, so these results compare the tested machines as configured rather than isolating one hardware component.

A CPU loop, a single AI request and several requests running at once are different workloads; use the result that matches the work you expect to do.

How did the GR1X perform in a sustained CPU test?

In StorageReview’s 30-minute Cinebench 2026 multi-core loop, the GR1X finished with a higher steady-state score than the two tested laptops. Its measured CPU temperatures were also lower during that run.

System and memorySteady-state scoreMean CPU temperaturePeak CPU temperature
ASUS ProArt GR1X, 64GB7,306 points83.3°C88.7°C
HP OmniBook Ultra 16, 128GB6,970 points97.4°C98.2°C
Dell XPS 16 Creator Edition, 64GB6,550 points97.3°C98.3°C

StorageReview reports that the GR1X did not throttle during this loop. Its single-core Cinebench 2026 score was 553 points, below the HP’s 561 and the Dell’s 565. The measured advantage is therefore clearest in this sustained multi-core run, not in single-core performance.

How does local AI output change with more requests?

For Qwen3.8 27B served through llama.cpp CUDA, the GR1X’s output rate rose from 13.1 tokens per second with one request to 40.0 with eight requests running concurrently. StorageReview described single-request results across its four-system comparison as within 6%.

SystemOutput at 1 requestOutput at 8 concurrent requests
ASUS ProArt GR1X, 64GB13.1 tokens/s40.0 tokens/s
HP OmniBook Ultra 16, 128GB12.6 tokens/s33.8 tokens/s
Dell XPS 16 Creator Edition, 64GB12.4 tokens/s31.9 tokens/s

At four concurrent requests, the GR1X produced 31.4 tokens per second. The MSI EdgeMesa N AI+ reached 39.8 tokens per second at eight requests, close to the GR1X’s 40.0. The results make concurrency an important part of the comparison: the desktop’s measured output separates more clearly from the tested laptops when several requests run together.

What do the other workloads show?

NotebookcheckReviews’ video shows benchmark software, QHD gaming, acoustic testing and a roughly 90GB local Qwen workflow on a 128GB/2TB GR1X.

Results outside the CPU loop and LLM test vary by task. In Blender 4.5’s CPU tests, the GR1X scored 215.6 samples per minute in Monster and 95.7 in Classroom. In Junkshop it reached 127.2 samples per minute, just below the HP’s 127.7.

For HandBrake x265 encoding, StorageReview measured 44.1 frames per second on the GR1X, compared with 40.6 on the HP and 38.8 on the Dell. Image-generation results were mixed: the GR1X produced 10.07 images per minute in FLUX.2 klein 4B Q4_0, below the HP’s 11.01 and the Dell’s 11.17. Its Z-Image Turbo 6B BF16 result was 4.17 images per minute, close to the HP’s 4.19.

StorageReview also measured a 10.94-second cold load for a 12.8GB model on the GR1X’s Gen4 SSD; a warm load took 3.68 seconds. The review attributed the longer cold-load time to the drive generation.

Notebookcheck’s earlier report characterized the GR1X’s graphics performance as around the GeForce RTX 5070 Laptop level. StorageReview’s results here compare specific workloads and systems rather than providing a direct RTX 5070 Laptop GPU result, so its figures do not quantify that earlier comparison.

NotebookcheckReviews separately demonstrates benchmarks, QHD gaming, acoustics and a local Qwen model workflow on a 128GB/2TB GR1X. Those demonstrations concern a different memory and storage configuration from StorageReview’s 64GB/1TB benchmark unit.