Beelink GTR9 Pro 495 Price: 192GB at $7,399, Ships From China in 35 Days
Beelink's GTR9 Pro 495 mini PC costs $4,799 with 128 GB and $7,399 with 192 GB, level with Minisforum's sold-out MS-S1 MAX-P495 as the highest 192 GB price we track. It ships from China in about 35 days, duties extra, with no LLM tests yet.

The Beelink GTR9 Pro 495, which Beelink lists at $7,399 with 192 GB of memory and a 2 TB SSD.
Image: Beelink
Source · Press image from the Beelink product page, used for news reporting.
The Beelink GTR9 Pro 495, built on the Ryzen AI Max+ 495, which AMD names the Ryzen AI Max+ PRO 495, is now on pre-order worldwide at $7,399 for 192 GB of memory and a 2 TB SSD, according to Beelink's store and a Notebookcheck report on Nov. 13, 2026.
The listing has three configurations: $4,799 for 128 GB and 2 TB, $7,399 for 192 GB and 2 TB, and $7,799 for 192 GB and 4 TB. The price is the same whichever plug a buyer picks (US, EU, UK, Australian or Japanese). Every order is a pre-order. Beelink's store says shipping takes about 35 days, every address is served from its warehouse in China, and the customer is responsible for any customs duties. No independent lab or outlet has yet published a large-language-model measurement of this machine, so this file sets out what Beelink states and what is still open.
Beelink GTR9 Pro 495 Specs: Memory, Ports and Power
The GTR9 Pro is a 180 x 180 x 90.8 mm aluminum mini PC weighing 2,338.8 g, by Beelink's spec table, with a built-in 230 W power supply rather than an external brick. Beelink lists 16 Zen 5 CPU cores and 32 threads at up to 5.2 GHz, Radeon 8065S integrated graphics with 40 compute units, and an XDNA 2 neural processing unit. It quotes 132 AI TOPS without naming the precision; AMD's PRO 495 page lists the same 132 as "Overall TOPS" and gives the NPU alone as up to 55 TOPS, also with no precision. Notebookcheck labels the 55 as INT8.
The memory is LPDDR5X-8533, soldered to the board as eight 24 GB packages in the 192 GB model, so it cannot be upgraded later. Storage sits in two M.2 2280 PCIe 4.0 x4 slots that take up to 8 TB each. The networking is the part aimed at local AI: two 10 Gb Ethernet ports and two 40 Gbps USB4 ports, which Beelink pitches for clustering several units. It ships with Windows 11 Pro and a three-year warranty. Beelink cools it with two fans and a vapor chamber and claims up to 140 W at 32 dB in its Full Performance Mode, a figure no one has tested yet; its spec graphic also shows a default TDP of 120 W.
Beelink's store title and spec table say "Ryzen AI Max+ 495", but a feature banner and the footnote of its thermal test (a 128 GB unit) say "Ryzen AI Max+ PRO 495". AMD's name, from its product page and its May 20, 2026 announcement, is Ryzen AI Max+ PRO 495, and AMD lists no non-PRO 495. Notebookcheck used the PRO name in September and dropped it on Nov. 13.
What Changed From the China Launch
This is the second chip in the GTR9 Pro chassis. The original, with the Ryzen AI Max+ 395 and up to 128 GB, was revealed in May 2025 and sold globally from August 2025, Notebookcheck reported. The 495 version first appeared in China, where Notebookcheck reported on Sept. 29, 2026 that it retailed for CNY 29,998 (about $4,472) with 128 GB and 2 TB, citing JD.com via ITHome, and that Beelink then offered no larger memory option. The same report said the power envelope stayed at 140 W with the 230 W supply, as on the 395 model.
The international listing adds what the China launch lacked: a 192 GB configuration, on AMD's 256-bit LPDDR5x interface.
How Unified Memory Sets Model Size
On these chips the processor cores and the graphics cores share one pool of LPDDR5X memory instead of a graphics card's own video memory. A model therefore fits only if the GPU may claim enough of that pool. AMD lists the Ryzen AI Max+ PRO 495 at up to 192 GB of unified memory, with up to 160 GB usable as dedicated graphics memory. Notebookcheck reported that the 128 GB Beelink version officially supports up to 96 GB.
Beelink's page does not say how much of the 192 GB its firmware gives the GPU, so buyers cannot yet tell whether the $7,399 model reaches AMD's 160 GB. The memory is soldered, so the $4,799 (128 GB) or $7,399 (192 GB) choice is final at purchase.
What 192 GB of Memory Means for Local AI Models
AMD says the platform runs 300B-parameter models at 4-bit. Framework, which sells a 192 GB desktop on the same AMD chip, says that configuration runs 300B-class models at 4-bit; both are the makers' claims. GMKtec claims up to 320B parameters offline for its EVO-X5 Pro, depending on model, software and quantization. Beelink makes no such claim. Its page names Qwen3.8-35B-A3B as an example of a model it can deploy locally and shows "4.24 tok/s" for it in LM Studio, a Beelink figure with no quantization, context, backend or memory size stated.
The second pitch is a team node. Two 10GbE ports and the built-in power supply make a box that sits on a shelf and serves several users, or links to a second unit. Whether that clustering works for splitting one model across two machines depends on software that nobody has yet shown running on this box.
GTR9 Pro 495 Price Against GMKtec, Bosgame and Framework
At $7,399, the GTR9 Pro is the most expensive 192 GB, 2 TB PRO 495-class box in the Voltgrid catalog, level with Minisforum's MS-S1 MAX-P495, which Minisforum lists at the same $7,399 but shows as sold out. GMKtec's EVO-X5 Pro lists at $6,799 with 2 TB and $7,099 with 4 TB, in stock on GMKtec's US store when read on Nov. 13. Bosgame's M5 MAX opened pre-orders on Nov. 3 at an introductory $6,199 for 192 GB and 2 TB, with local shipping in the US, UK and EU but no firm ship date, Notebookcheck reported. The Framework Desktop with 192 GB cost $6,799 as a DIY pre-order without storage on Nov. 3, with its third batch due in December. AMD's own 192 GB Ryzen AI Halo was released the week of Oct. 12 at $7,299.99 for the Linux version, Tom's Hardware reported; our review of the 128 GB Ryzen AI Halo covers the platform's software.
On list prices, by Voltgrid arithmetic, Beelink's 192 GB, 2 TB model works out to $7,399 divided by 192 GB, or $38.54 per GB, against $35.41 for GMKtec's $6,799 model and $32.29 for Bosgame's $6,199 offer, each with a 2 TB SSD included; Framework is left out because its $6,799 has no storage. The 128 GB model is $4,799 divided by 128 GB, or $37.49 per GB. Our Nov. 12 price comparison did not include the Beelink. On paper, the premium over GMKtec buys an internal power supply where the EVO-X5 Pro uses a 330 W external adapter, plus dual 10GbE; Bosgame's cheaper M5 MAX also has dual 10GbE, TechRadar reported.
On list prices, Beelink's 192 GB model costs more per GB than GMKtec's or Bosgame's (Voltgrid arithmetic; Bosgame's price is an introductory pre-order).
Illustration (AI-generated)
Shipping changes the real price. Beelink ships every order from China and leaves duties to the buyer without saying what they will be. A US or EU buyer should budget for an import charge that the $7,399 does not include, and for a wait: Beelink's 35-day shipping estimate points to mid-to-late December; the page does not say whether transit or customs time is included.
What Is Unknown Until Reviews Land
Speed is the largest gap. Beelink does not state memory bandwidth, and its LPDDR5X-8533 rating alone does not settle what the box delivers. Framework and GMKtec state 273 GB/s for their PRO 495 machines, GMKtec as "up to", but no one has measured this one. No lab or outlet has published a tokens-per-second figure for this machine; Beelink's own 4.24 tok/s on Qwen3.8-35B-A3B in LM Studio is the only number, and it comes with no quantization or context. The nearest independent data point is from a different machine: TechRadar's Nov. 4 review of the GMKtec EVO-X5 Pro measured Qwen3-Coder-Next at 45 tokens per second in the Lemonade server, without stating the quantization.
Also open are the GPU memory limit on the 192 GB unit, whether the 140 W mode holds under a long inference load and at what noise, how the 230 W internal supply copes with sustained load, the Linux and ROCm support Beelink offers for a Windows-first box, and the precision behind its 132 TOPS.
Beelink's 35-day estimate points to mid-to-late December for units ordered now. ServeTheHome reviewed the original Ryzen AI Max+ 395 GTR9 Pro, but no outlet has announced a review of the 495 model, and Beelink has not said when review units will go out. Until a lab publishes a run on a named model and quantization, the GTR9 Pro's case rests on its ports and its price, not on measured speed.
How it compares
Full table| Product | Memory | Bandwidth | AI compute | Measured speed | Price |
|---|---|---|---|---|---|
| ASUS ExpertCenter Pro ET900N G3ASUS (NVIDIA DGX Station GB300 platform) | 748 GB coherent: 252 GB HBM3e (GPU) + 496 GB LPDDR5X (CPU)[1] | 7.1 TB/s (GPU HBM3e); 396 GB/s (CPU LPDDR5X)[1] | up to 20 PFLOPS AI performance (ASUS gives no precision or sparsity label) (sparse)[2] | GPT-OSS-120B MXFP4, vLLM, 512/512, 1-128 streams: 9,280 aggregate output tok/s at 128 streams; 5,023 on 8,192/1,024 (StorageReview, ASUS-supplied unit, no add-in GPU)[3] | $111,999.99 at Central Computers, US (ET900N G3-2SSD-V2, read Nov 8, 2026); no ASUS list price[4] |
| HP ZGX Fury AI StationHP | 252 GB HBM3e GPU + 496 GB LPDDR5x CPU (4x 128 GB SOCAMM); 748 GB unified[5] | 7.1 TB/s (GPU HBM3e); 396 GB/s (CPU LPDDR5x)[5] | up to 20 petaFLOPS FP4 (HP gives no sparsity label) (sparse)[5] | Two-unit cluster with an ASUS ET900N G3 over two 400G DACs: GLM-5.3 NVFP4 5,018 output tok/s at 256 streams, 512/512, vLLM 0.27.1 TP2 with MTP (StorageReview); no single-unit HP result[6] | $132,000.00 on HP Store US (ZGX Fury G1n, 3-yr warranty, no graphics card, 4x 2 TB SED SSDs; Add to cart; read Nov 11, 2026)[7] |
| NVIDIA DGX StationNVIDIA (systems built by ASUS, Dell, GIGABYTE, HP, MSI, Supermicro and Exxact) | 748 GB coherent: 252 GB HBM3e (GPU) + 496 GB LPDDR5X (CPU)[8] | GPU HBM3e 7.1 TB/s; CPU LPDDR5X 396 GB/s; NVLink-C2C 900 GB/s between them[8] | 15 PFLOPS FP4 (without sparsity)[8] | gpt-oss-120b MXFP4, vLLM, 512/512 chat workload: 9,280 output tok/s at 128 streams (MSI WS300 9,294), single GB300 (StorageReview, ASUS ExpertCenter Pro ET900N G3)[3] | No NVIDIA list price; lowest US partner listing read Nov 4, 2026: Exxact Valence VWS-158270643 from $98,450 (Tom's Hardware found $94,930 in Aug); Dell Pro Precision with GB300 lists $160,195[9] |
| Supermicro Super AI Station (ARS-511GD-NB-LCC)Supermicro (NVIDIA DGX Station platform) | 748 GB total coherent: 252 GB HBM3e GPU + 496 GB LPDDR5X system[10] | 7.1 TB/s (GPU HBM3E); 396 GB/s (CPU LPDDR5X)[11] | up to 20 PFLOPS FP4 (Supermicro gives no sparsity label; ServeTheHome describes it as sparse FP4)[10] | Qwen3.8 Flash Next NVFP4, vLLM, tool-loop 4,096-in/512-out: several scenarios exceeded 3,000 output tok/s at 32 concurrent users, still rising (ServeTheHome; RTX PRO 2000 installed)[12] | $92,887.47 before tax and shipping on Supermicro's online store (ARS-511GD-NB-LCC-01-G2); about $90,101.05 with a 3% new-customer discount (TechRadar, Sep 1, 2026)[13] |
| NVIDIA RTX PRO 6000 Blackwell Workstation EditionNVIDIA (sold through PNY, TD SYNNEX and workstation makers) | 96 GB[14] | 1,792 GB/s[14] | 4,000 AI TOPS (FP4, with sparsity) (sparse)[14] | gpt-oss-120b MXFP4, stock vLLM, 512/512 chat workload: 3,064 output tok/s at 128 streams (1,169 on 8,192/1,024 prompts); Server Edition at 600W (StorageReview)[15] | $16,000 on NVIDIA's US marketplace (Workstation Edition), listed out of stock[16] |
| Apple Mac Studio (M5 Ultra)Apple | up to 512 GB unified (96 GB on 30-core M5 Ultra; 256 GB or 512 GB on 36-core/80-core M5 Ultra)[17] | 1.2 TB/s[17] | — | GLM-5.3-Flash MLX 4-bit with MTP on TensorFold, ISL 2048/OSL 512: 94.0 tok/s at 0 context, 96.5 at 128K, 79.4 at 512K; TTFT 1.1 s at 0, 66.3 s at 128K (Tom's Hardware, 256 GB M5 Ultra)[18] | $22,299 minimum with 512 GB ($6,799 36-core CPU/80-core GPU M5 Ultra config plus $15,500 memory upgrade; Oct 28, 2026)[19] |
| NVIDIA RTX Spark (N1X) PCsNVIDIA (MediaTek collaborated on the custom CPU design) | Up to 128 GB LPDDR5X (20-core CPU / 6,144-core GPU version); up to 64 GB (18-core / 5,120-core laptop version)[20] | Up to 300 GB/s: LPDDR5X-9400 on a 256-bit bus (NVIDIA's product page lists no bandwidth)[21] | — | gpt-oss:20b via Ollama in a Codex CLI coding agent, 64K context: 58.5 tok/s generation, 3 of 4 tasks passed (Dell XPS 16, 20-core N1X, 64 GB; StorageReview)[22] | Laptops from $2,599; pricing and configurations vary by region[23] |
| AMD Ryzen AI Halo (Ryzen AI Max+ PRO 495, 192 GB)AMD | 192 GB LPDDR5X (256-bit)[24] | 273 GB/s[24] | — | — | $7,299.99 for the Linux version (released the week of Oct 12, 2026); Windows version coming soon[24] |
Show the other 29 machinesHide the other 29 machines
| Product | Memory | Bandwidth | AI compute | Measured speed | Price |
|---|---|---|---|---|---|
| ASUS Ascent GX10ASUS | 64 GB or 128 GB LPDDR5X-8533 unified[25] | 273 GB/s (8533 MT/s), as listed by StorageReview; ASUS's spec page states no bandwidth[26] | Up to 1 PFLOP FP4 (GB10; NVIDIA's footnote: theoretical, using the sparsity feature). ASUS's pages say 1 petaFLOP / FP4 with no sparsity label (sparse)[27] | gpt-oss-120b, vLLM online serving, equal ISL/OSL: 70 tok/s at batch 1 rising to 680 tok/s aggregate at batch 64 (StorageReview)[26] | $5,999 (128 GB, 1 TB, GX10-GG0010BN, ASUS eShop US, Oct 30, 2026)[28] |
| Dell Pro Precision with GB10 (model FCM1253; launched as Dell Pro Max with GB10)Dell | 128 GB LPDDR5x up to 8533 MT/s onboard (64 GB option also listed)[29] | 273 GB/s (128 GB LPDDR5x, 256-bit)[30] | — | gpt-oss-120b, vLLM online serving, equal ISL/OSL: 68.57 tok/s at batch 1 rising to 670.83 tok/s aggregate at batch 64 (StorageReview)[31] | $7,901.00 (default FCM1253, 128 GB, 2 TB Gen4 QLC); 4 TB TLC SED SSD option +$2,450.00; 64 GB option listed at a $2,150.00 difference (Dell.com US, read Nov 10, 2026)[32] |
| Framework Desktop (Ryzen AI Max+ PRO 495, 192 GB)Framework | 192 GB LPDDR5X-8533 soldered[33] | 273 GB/s (256-bit)[33] | — | — | $6,799 (Max+ PRO 495 192 GB DIY Edition pre-order, Nov 3, 2026; storage not included)[34] |
| GIGABYTE AI TOP ATOMGIGABYTE | 128 GB unified system memory[35] | 273 GB/s (LPDDR5x-8533, 256-bit)[36] | — | gpt-oss-120b, vLLM online serving, equal ISL/OSL: 60.71 tok/s at batch 1 rising to 649.62 tok/s aggregate at batch 64 (StorageReview)[36] | $7,499.99 (128 GB, 4 TB PCIe 4.0) and $9,405.63 (128 GB, 4 TB PCIe 5.0; more options from $7,999.99); 128 GB from $5,999.99 (1 TB) and $6,999.99 (2 TB) at Newegg US, read Nov 9, 2026[37] |
| GMKtec EVO-X5 Pro AI Mini PCGMKtec | 192 GB LPDDR5X-8533 onboard (8 x 24 GB, non-upgradeable); up to 160 GB allocatable as VRAM[38] | Up to 273 GB/s[39] | — | GMKtec Laboratory table: GLM-5.3-Flash UD-IQ4_XS (146.05 GB) decode 17.79 tok/s, prefill 70.18, TTFT 14.71 s; Qwen3.8-Flash-Next ROCmFP4-FAST (93.48 GB) decode 45.08 tok/s[38] | $6,799 (192 GB/2 TB) and $7,099 (192 GB/4 TB), all variants available, GMKtec US store, read Nov 9, 2026[38] |
| HP ZGX Nano AI Station (G1n)HP | 128 GB coherent unified system memory[40] | 273 GB/s (128 GB LPDDR5x)[41] | GB10: up to 1 PFLOP FP4, theoretical, using the sparsity feature; HP and reviewers list 1,000 TOPS FP4 (sparse)[27] | gpt-oss-120b, vLLM online serving, equal ISL/OSL: 69.6 tok/s at batch 1 rising to 722.9 tok/s aggregate at batch 64 (StorageReview)[42] | $7,399.00 at the HP Store for the 4 TB model; $4,759 on Amazon (Notebookcheck, review updated Oct 5, 2026)[43] |
| Lenovo ThinkStation PGXLenovo | 128 GB LPDDR5X unified (256-bit)[44] | 273 GB/s[44] | — | — | US from $5,079 (1 TB base model) on Lenovo's site, per TechRadar's review (Jun 9, 2026); UK from GBP 5,200[45] |
| MSI EdgeMesa N AI+ (128 GB, X-005US)MSI | 128 GB LPDDR5X unified system memory[46] | 273 GB/s[46] | — | — | $7,999.00 (128 GB, 4 TB, X-005US, MSI US store, read Nov 11, 2026)[46] |
| MSI EdgeXpert (MS-C931)MSI | 64 GB or 128 GB LPDDR5x unified (256-bit)[47] | 273 GB/s[47] | 1 PFLOPS FP4 (sparse)[47] | — | No list price; MSI sells via quote (Get a quote)[47] |
| NVIDIA DGX SparkNVIDIA (partner GB10 systems from Acer, ASUS, Dell, GIGABYTE, HP, Lenovo and MSI) | 128 GB; a 64 GB configuration is sold only through OEM partners[27] | 273 GB/s[27] | Up to 1 PFLOP FP4 (theoretical, using the sparsity feature) (sparse)[27] | Qwen 3.8 27B (dense) with multi-token prediction: decode of about 20 tokens/s, with time-to-first-token staying interactive out to the full 262K context.[48] | $6,950.00 (128 GB, 4 TB) on NVIDIA's enterprise marketplace with an Add To Cart button in a copy cached 07:59 UTC on 2026-10-31; stock unverified (listed out of stock on 2026-10-21); 64 GB partner systems from $4,999[49] |
| NVIDIA DGX Spark 64 GBNVIDIA (sold by Acer, ASUS, Dell, Gigabyte, HP and MSI) | 64 GB LPDDR5X unified (two units pool 128 GB over ConnectX-7)[50] | 273 GB/s (256-bit; NVIDIA spec table covers both 64 GB and 128 GB)[27] | 1 PFLOPS FP4 (sparse; NVIDIA spec table covers both memory configurations)[27] | — | from $4,999 (partner systems, Oct 2026; no Founders Edition)[50] |
| Sipeed MaixStationSipeed | 128 GB LPDDR5X (256-bit)[24] | 273 GB/s (Sipeed claim)[24] | 256 TOPS FP4 (native FP4, no sparsity; Axera/Sipeed claim)[24] | — | about $3,999 expected for 128 GB, with Kickstarter launch discounts (Sipeed to Tom's Hardware, Oct 2026)[24] |
| AMD Ryzen AI Halo Developer PlatformAMD | 128 GB LPDDR5x-8000 unified[51] | 256 GB/s[51] | — | gpt-oss-120b, vLLM online serving, equal ISL/OSL 256/256: 222 tok/s aggregate at batch 64 vs 701 on DGX Spark (StorageReview)[52] | $3,999 with 2 TB SSD (Jul 2026)[52] |
| Framework Desktop (Ryzen AI Max+ 395)Framework | 128 GB LPDDR5x-8000 soldered (32 GB and 64 GB options)[53] | 256 GB/s[54] | — | gpt-oss-120b MXFP4 in LM Studio (out-of-the-box setup): roughly 45 tok/s (ServeTheHome; Ryzen AI Max+ 395, 128 GB, Windows 11 Pro)[55] | $3,449 (DIY Edition, Max+ 395 with 128 GB, system only; out of stock, Nov 3, 2026)[56] |
| Acer Veriton GN100 AI Mini WorkstationAcer | 128 GB LPDDR5x coherent unified[57] | — | — | gpt-oss-120b, vLLM, total token throughput: prefill heavy 303.61 to 2,777.56 tok/s; decode heavy 38.38 to 292.41 tok/s (batch 1 to 64, StorageReview)[58] | from $3,999 in North America (announced Sep 3, 2025; prices vary by region)[57] |
| ASRock Industrial AI BOX-A395ASRock Industrial | 128 GB LPDDR5X-8000 soldered[59] | — | — | AgentSTH V7, 32 threads: 15% ahead of the next-closest Ryzen AI Max+ 395 box (single-core within 1%); estimated SPEC CPU2026 int rate: top of a 44% spread across four 395 systems; measured by ServeTheHome, Ubuntu 26.04[60] | — |
| Beelink GTR9 Pro (Ryzen AI Max+ 495, 128 GB)Beelink | 128 GB LPDDR5X-8533 unified (soldered)[61] | — | 132 AI TOPS on Beelink's page; AMD lists the PRO 495 at Overall TOPS up to 132 and NPU TOPS up to 55; no precision stated by either[61] | 4.24 tok/s, LM Studio, Qwen3.8-35B-A3B (Beelink's own figure on its store page; quantization, context, backend and RAM size not stated)[61] | $4,799 (128 GB/2 TB), Beelink store, pre-order (read Nov 13, 2026)[61] |
| Beelink GTR9 Pro (Ryzen AI Max+ 495, 192 GB)Beelink | 192 GB LPDDR5X-8533 unified (soldered)[61] | — | 132 AI TOPS on Beelink's page; AMD lists the PRO 495 at Overall TOPS up to 132 and NPU TOPS up to 55; no precision stated by either[61] | 4.24 tok/s, LM Studio, Qwen3.8-35B-A3B (Beelink's own figure on its store page; quantization, context, backend and RAM size not stated)[61] | $7,399 (192 GB/2 TB) or $7,799 (192 GB/4 TB), Beelink store, pre-order (read Nov 13, 2026)[61] |
| Bosgame M5 MAX AI Mini PCBosgame | 192 GB unified LPDDR5 (soldered); up to 160 GB allocatable to the GPU[62] | — | — | — | $6,199 introductory first-batch pre-order price (192 GB, 2 TB PCIe 4.0 SSD), from Nov 3, 2026, until the first allocation sells out[62] |
| Dell Pro Precision with GB300 (model FCT6263; announced as Dell Pro Max with GB300)Dell | 252 GB HBM3e GPU + 496 GB LPDDR5X CPU[63] | — | — | — | $160,195.00 (Dell.com US default config FCT6263, read Nov 8, 2026)[63] |
| Gigabyte AI TOP Aero | up to 128 GB LPDDR5X unified[64] | — | — | — | — |
| GMKtec EVO-X2 AI Mini PCGMKtec | 64 GB or 128 GB onboard LPDDR5X-8000 (non-upgradeable)[65] | — | — | Llama 3.3 70B Q6_K Instruct: 3.7-3.8 tok/s generation fully on the Radeon 8060S (about 2 tok/s CPU-only); Qwen3-32B 10.4-10.5 tok/s; 128 GB unit, short prompt, tool not named (ServeTheHome)[66] | $3,499.99 (128 GB/1 TB) and $3,649.99 (128 GB/2 TB), each against a $3,999.99 compare-at price (GMKtec US store, read Nov 11, 2026)[65] |
| HP Z2 Mini G1a WorkstationHP | up to 128 GB unified, up to 96 GB assignable to the GPU, up to 8000 MT/s[67] | — | up to 50 NPU TOPS[67] | gpt-oss-120b GGUF (63.39 GB, full GPU offload) in LM Studio: close to 40 tok/s (StorageReview)[68] | $4,781 list for StorageReview's review configuration; about $3,300 street (B&H top spec with 2 TB SSD at $3,342.65, Aug 2025)[68] |
| Microsoft Surface RTX Spark Dev BoxMicrosoft | 128 GB unified memory[69] | — | up to 1 PFLOP of AI compute (Microsoft; precision and sparsity not stated on the page) (sparse)[69] | — | From $5,999.99 (Microsoft Store US, pre-order, read Nov 1, 2026)[69] |
| MSI EdgeMesa N AI+ (64 GB) | — | — | — | — | from $4,699 in the US (64 GB soldered LPDDR5X, 1 TB PCIe Gen 5)[70] |
| MSI XpertStation WS300MSI (NVIDIA DGX Station platform) | 252 GB HBM3e GPU + 496 GB LPDDR5X CPU (748 GB coherent)[71] | — | — | StorageReview vLLM: gpt-oss-120b MXFP4 9,294 output tok/s at 128 streams (512/512), 5,038 on 8,192/1,024; MiniMax M3 NVFP4 1,041 at 32 streams with EAGLE3-GQA at a forced acceptance length of 3 (best-case)[71] | $99,999.00 'Start From' at Newegg US (model WS300T60L, 2x 2 TB NVMe RAID1), sold by Newegg as a request-a-quote listing (read Nov 4, 2026)[72] |
| Tenstorrent TT-QuietBox 2 (Blackhole)Tenstorrent | 128 GB GDDR6 on the accelerators + 256 GB DDR5-5600 system memory[73] | — | — | — | $9,999 (Oct 30, 2026)[73] |
| WiCi One (RTX 5060 Ti edition)WiCi | 16 GB GDDR7 VRAM; 1 TB NVMe 4 SSD for TurboStream weight paging (WiCi's claim: NVMe used as working VRAM). System RAM not stated[74] | — | — | — | $2,599 list (WiCi shop, Oct 2026); a $600-off code (WICIWELCOME) is applied at checkout. Early-access price was $1,999 (Notebookcheck, Engadget, Sep 2026)[74] |
| WiCi One (RTX 5090 edition)WiCi | 32 GB GDDR7 VRAM; 4 TB NVMe 5 SSD for TurboStream weight paging (WiCi's claim: NVMe used as working VRAM). System RAM not stated[74] | — | — | — | $9,999 list (WiCi shop, Oct 2026); a $600-off code (WICIWELCOME) is applied at checkout[74] |
Italic values are reported by an outlet or measured by an independent lab; the rest come from the maker. Measured speeds use different models, quantizations and tools, so read each with its source rather than against the others. Voltgrid News does not test hardware itself.
Sources (74)
- [1] ASUS
- [2] ASUS
- [3] StorageReview
- [4] Central Computers
- [5] HP
- [6] StorageReview
- [7] HP
- [8] NVIDIA
- [9] Exxact
- [10] Supermicro
- [11] ServeTheHome
- [12] ServeTheHome
- [13] TechRadar
- [14] NVIDIA
- [15] StorageReview
- [16] NVIDIA
- [17] Apple
- [18] Tom's Hardware
- [19] MacRumors
- [20] NVIDIA
- [21] ServeTheHome
- [22] StorageReview
- [23] NVIDIA
- [24] Tom's Hardware
- [25] ASUS
- [26] StorageReview
- [27] NVIDIA
- [28] ASUS eShop
- [29] Dell
- [30] HotHardware
- [31] StorageReview
- [32] Dell
- [33] Framework Blog
- [34] Framework
- [35] GIGABYTE (PR Newswire)
- [36] StorageReview
- [37] Newegg
- [38] GMKtec
- [39] GMKtec
- [40] HP
- [41] HotHardware
- [42] StorageReview
- [43] Notebookcheck
- [44] Lenovo PSREF
- [45] TechRadar
- [46] MSI
- [47] MSI
- [48] Tom's Hardware
- [49] NVIDIA Marketplace
- [50] NVIDIA Blog
- [51] AMD
- [52] StorageReview
- [53] Framework
- [54] Framework Blog
- [55] ServeTheHome
- [56] Framework
- [57] Acer
- [58] StorageReview
- [59] ServeTheHome
- [60] ServeTheHome
- [61] Beelink
- [62] Notebookcheck
- [63] Dell
- [64] Notebookcheck
- [65] GMKtec
- [66] ServeTheHome
- [67] HP
- [68] StorageReview
- [69] Microsoft
- [70] Notebookcheck
- [71] StorageReview
- [72] Newegg
- [73] Tenstorrent
- [74] WiCi
Sources
- Desk
- Local AI Desk
- Editor
- Ege Erten
- Published
- Sources
- 2 cited
- Standards
- AI-assisted
