VoltgridNews

Corsair FlexPrime V20C: $3,999.99 Arc Pro B70 Tower Runs 27B at 19.8 tok/s

Corsair FlexPrime V20C: StorageReview, the only outlet with measurements as of Nov. 15, got 19.8 tok/s on Qwen3.8 27B Q4_K_M at one request; Corsair rates the 32 GB card at 608 GB/s.

Local AI Desk

6 min read

Corsair FlexPrime V20C workstation tower in black, side mesh panel and Corsair PRO badge

The Corsair FlexPrime V20C, a $3,999.99 workstation built around the Intel Arc Pro B70 (Corsair product image).

Image: Corsair

Source · Press image from the Corsair product page, used for news reporting.

The Corsair FlexPrime V20C is a $3,999.99 Linux workstation built around Intel's Arc Pro B70, a graphics card with 32 GB of GDDR6 that Corsair rates at 608 GB/s. StorageReview, the only outlet we found publishing measurements by Nov. 15, ran Qwen3.8 27B at Q4_K_M in the llama.cpp server at 19.8 tokens per second for one request.

That review, published on Nov. 15, is one data set from one lab. It is enough to say what the machine is and where it looks strong, but not yet enough for a full verdict. The figures below name who measured them and under which settings, and the open questions are listed at the end.

What the FlexPrime V20C Is

The V20C (SKU CSW-9040004-NA) is a fixed configuration in Corsair's CORSAIR PRO line of AI workstations, which StorageReview says the company introduced in May 2026. Corsair's US store lists it at $3,999.99, in stock, with an estimated delivery of Nov. 17 to Nov. 19 when read on Nov. 15. The price covers an Intel Core Ultra 7 270K Plus with 24 cores, the Arc Pro B70, 32 GB of DDR5 as two 16 GB modules, a 1 TB M.2 NVMe SSD, an 850 W power supply, a 360 mm liquid CPU cooler and Ubuntu 24.04 LTS, with a three-year warranty, according to Corsair's product page.

Corsair states 367 INT8 TOPS for the graphics card, without a dense or sparse label (Intel's own specification lists 367 as dense), and 256 XMX AI engines. StorageReview measured the box at 202 x 451 x 286 mm and 14 kg, a micro-ATX tower on an Intel Z890 board built from standard desktop parts, so the memory, storage, card and power supply can be swapped later.

The new part is the memory arrangement at this price. Its closest rivals at this price, AMD's Ryzen AI Halo and NVIDIA's GB10 boxes, share one pool of unified memory between processor and graphics. The V20C instead gives the model its own 32 GB on a discrete card. Intel launched the B70 at $949, StorageReview noted, which is how a $4,000 tower reaches 32 GB of graphics memory.

Bandwidth is the trade. Token generation for a model that fits mostly waits on how fast the weights stream out of memory. The B70's 608 GB/s compares with 256 GB/s that AMD lists for its 128 GB Ryzen AI Halo and 273 GB/s that NVIDIA lists for the GB10 in the DGX Spark. The comparison is not like for like: the 608 GB/s belongs to 32 GB of GDDR6 on one card, while the 256 GB/s and 273 GB/s feed 128 GB unified LPDDR5X pools shared with the processor. Capacity runs the other way: those boxes carry four times the card's 32 GB.

The second change is the software. Corsair ships an Ubuntu image with the GPU driver and a compute runtime, and advertises an optional AI stack with Docker, monitoring and Python. Its setup guide serves Qwen3.8-27B with llama.cpp at a 65,536-token context, StorageReview reported. The lab replaced the factory image with Ubuntu 26.04.1 LTS before its runs, so the shipped image itself was not measured.

Which Local AI Models It Can Run

The machine targets a developer or small team that wants to serve a model of about 27B parameters locally, through llama.cpp or Intel's OpenVINO, on an x86 tower that also compiles code quickly. StorageReview's numbers support that use. It found 31.9 GB of the card usable, and Qwen3.8 27B at Q4_K_M needs about 30 GB for its weights and one request's cache, which the 24 GB RTX PRO 4000 SFF in its comparison could not load.

On that model, StorageReview measured 19.8 tokens per second at one request, 33.5 at four and 35.4 at eight, with a first token after 9,005 ms on 8,192-token prompts. On the smaller Ministral 3 8B, with 4,096-token prompts and 1,024-token outputs on llama.cpp's SYCL backend, it measured 63.7 tokens per second at one request and 141.8 at four. MLPerf Client v2.0 on OpenVINO returned 105.4 tokens per second on Llama 3.1 8B. On the CPU side, the 270K Plus built LLVM in 136 seconds, the fastest in StorageReview's set.

Corsair's product page claims about 55 tokens per second on Qwen3.8-27B. It does not state the quantization, the inference engine, the context length or the number of requests behind that figure, and StorageReview called it directional. It is Corsair's claim, not a measurement, and it cannot be set beside StorageReview's 19.8.

FlexPrime V20C vs GB10 and Ryzen AI Halo Boxes

StorageReview ran Qwen3.8 27B in the llama.cpp server on two rivals, each on its own vendor's backend (the lab names Q4_K_M for the model's memory need and does not restate the quantization per box): SYCL on the V20C, CUDA on the Dell Pro Max with GB10 (listed on Dell.com now as the Dell Pro Precision with GB10), which it tested with 128 GB and a 4 TB drive, and ROCm on the AMD Ryzen AI Halo. At one request on Qwen3.8 27B, the GB10 box reached 11.2 tokens per second and the Halo 9.1, against 19.8 on the V20C. At eight requests the GB10 reached 35.6, level with the V20C's 35.4, while the Halo reached 20.1. The single-user lead is in generation, since time to first token was 9.0 seconds on the V20C and 9.2 on the GB10.

StorageReview's one-request Qwen3.8 27B speeds in llama.cpp (SYCL, CUDA, ROCm); memory and bandwidth are each maker's specification.

Graphic: Voltgrid News

The headline speeds in the comparison table under this story are a different test. For those rivals they are gpt-oss-120b runs: in StorageReview's July review of the Halo, gpt-oss-120b in vLLM with 256-token prompts and outputs reached an aggregate 222 tokens per second at batch 64 on the Halo, against 701 on the DGX Spark. StorageReview did not run models that need 38 GB or more on the V20C, so no gpt-oss-120b figure exists for it, and its Qwen3.8 27B results are not comparable with those entries.

Price is close on paper. AMD's Halo was reported at $3,999 with a 2 TB SSD in July. NVIDIA says 64 GB DGX Spark partner systems start at $4,999, with 273 GB/s and twice the V20C's 32 GB, as one unified LPDDR5X pool shared with the processor rather than dedicated graphics memory. The Dell GB10 box that StorageReview compared had a 4 TB drive; the lab cites $4,061.34 for a 1 TB model at the time of its earlier review and says the price is much higher now, and Dell.com listed the default 128 GB, 2 TB configuration at $7,901.00 on Nov. 10.

Why Memory Bandwidth Sets Local Token Speed

A language model writes one token at a time, and each token means reading the model's weights from memory again. A single user's speed therefore follows how fast memory delivers data, not only how much compute the chip has, which is why the card's 608 GB/s shows up as the V20C's lead at one request.

Capacity decides what loads at all. The 31.9 GB that StorageReview found usable holds Qwen3.8 27B at Q4_K_M, but not the workloads of 38 GB or more that the 128 GB boxes ran. Once several requests share the card, compute and batching weigh more than bandwidth, and the GB10 box drew level at eight requests.

Known Issues and Open Questions

The weak spots StorageReview found are mostly in software. Ollama's own Intel path loaded no model layers onto the B70, so the lab switched Ollama to Vulkan, where the V20C generated 32.2 tokens per second on gpt-oss:20b in a coding-agent task against 53.9 on the GB10, which ran CUDA. One Vulkan run ended in a device-lost error before passing on a rerun. Image generation on stable-diffusion.cpp reached 4.26 FLUX.2 klein 4B Q4_0 images per minute on Vulkan against 11.46 on the RTX PRO 4000 SFF on CUDA; the GB10 on Vulkan reached 8.29. Fine-tuning was not run, because the suite's PyTorch builds cover only CUDA and ROCm, so Corsair's fine-tuning pitch is unmeasured.

The review unit also differed from the listing. Corsair shipped StorageReview 64 GB of DDR5 instead of the listed 32 GB, and the lab measured STREAM Triad at 68 GB/s from the two DDR5 channels, which matters for anything that spills off the card. StorageReview reported irregular fan ramping and an SSD that reached 80.8 C. It gave no wall-power figure under load, and it did not test vLLM.

Several questions remain open: wall power, whether the shipped Corsair image behaves like the lab's install, how vLLM and Intel's serving stack run on this box, and what settings produced Corsair's 55 tokens per second.

No second independent measurement of the V20C had turned up in our search by Nov. 15. Corsair's store gave first deliveries as Nov. 17 to Nov. 19. Until another lab reports, StorageReview's figures on the 64 GB review unit are the only independent evidence, and a second lab's numbers are needed before the machine can be judged on more than one data set.

How it compares

Full table
Corsair FlexPrime V20C: $3,999.99 Arc Pro B70 Tower Runs 27B at 19.8 tok/s: how the machine compares with other local AI hardware
ProductMemoryBandwidthAI computeMeasured speedPrice
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 StationHP252 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]
MSI XpertStation WS300MSI (NVIDIA DGX Station platform)252 GB HBM3e GPU + 496 GB LPDDR5X CPU (748 GB coherent)[8]7.1 TB/s (GPU HBM3e); 396 GB/s CPU LPDDR5X[8]—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)[8]$99,999.00 'Start From' at Newegg US (model WS300T60L, 2x 2 TB NVMe RAID1), request-a-quote listing sold by Newegg (read Nov 15, 2026)[9]
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)[10]GPU HBM3e 7.1 TB/s; CPU LPDDR5X 396 GB/s; NVLink-C2C 900 GB/s between them[10]15 PFLOPS FP4 (without sparsity)[10]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[11]
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[12]7.1 TB/s (GPU HBM3E); 396 GB/s (CPU LPDDR5X)[13]up to 20 PFLOPS FP4 (Supermicro gives no sparsity label; ServeTheHome describes it as sparse FP4)[12]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)[14]$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)[15]
NVIDIA RTX PRO 6000 Blackwell Workstation EditionNVIDIA (sold through PNY, TD SYNNEX and workstation makers)96 GB[16]1,792 GB/s[16]4,000 AI TOPS (FP4, with sparsity) (sparse)[16]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)[17]$16,000 on NVIDIA's US marketplace (Workstation Edition), listed out of stock[18]
Apple Mac Studio (M5 Ultra)Appleup to 512 GB unified (96 GB on 30-core M5 Ultra; 256 GB or 512 GB on 36-core/80-core M5 Ultra)[19]1.2 TB/s[19]—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)[20]$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)[21]
Corsair FlexPrime V20C Workstation (CSW-9040004-NA)Corsair32 GB GDDR6 on the Arc Pro B70 GPU + 32 GB (2x 16 GB) DDR5 system memory[22]608 GB/s (GPU GDDR6)[22]—Qwen3.8 27B Q4_K_M, llama.cpp server: 19.8 tok/s at 1 request, 33.5 at 4, 35.4 at 8; TTFT 9,005 ms on 8,192-token prompt (StorageReview)[23]$3,999.99 on Corsair US store (32 GB DDR5, 1 TB SSD; read Nov 15, 2026)[22]
Show the other 32 machines
Corsair FlexPrime V20C: $3,999.99 Arc Pro B70 Tower Runs 27B at 19.8 tok/s: how the machine compares with other local AI hardware: the other machines
ProductMemoryBandwidthAI computeMeasured speedPrice
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)[24]Up to 300 GB/s: LPDDR5X-9400 on a 256-bit bus (NVIDIA's product page lists no bandwidth)[25]—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)[26]Laptops from $2,599; pricing and configurations vary by region[27]
AMD Ryzen AI Halo (Ryzen AI Max+ PRO 495, 192 GB)AMD192 GB LPDDR5X (256-bit)[28]273 GB/s[28]——$7,299.99 for the Linux version (released the week of Oct 12, 2026); Windows version coming soon[28]
ASUS Ascent GX10ASUS64 GB or 128 GB LPDDR5X-8533 unified[29]273 GB/s (8533 MT/s), as listed by StorageReview; ASUS's spec page states no bandwidth[30]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)[31]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)[30]$5,999 (128 GB, 1 TB, GX10-GG0010BN, ASUS eShop US, Oct 30, 2026)[32]
Dell Pro Precision with GB10 (model FCM1253; launched as Dell Pro Max with GB10)Dell128 GB LPDDR5x up to 8533 MT/s onboard (64 GB option also listed)[33]273 GB/s (128 GB LPDDR5x, 256-bit)[34]—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)[35]$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)[36]
Framework Desktop (Ryzen AI Max+ PRO 495, 192 GB)Framework192 GB LPDDR5X-8533 soldered[37]273 GB/s (256-bit)[37]——$6,799 (Max+ PRO 495 192 GB DIY Edition pre-order, Nov 3, 2026; storage not included)[38]
GIGABYTE AI TOP ATOMGIGABYTE128 GB unified system memory[39]273 GB/s (LPDDR5x-8533, 256-bit)[40]—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)[40]$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[41]
GMKtec EVO-X5 Pro AI Mini PCGMKtec192 GB LPDDR5X-8533 onboard (8 x 24 GB, non-upgradeable); up to 160 GB allocatable as VRAM[42]Up to 273 GB/s[43]—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[42]$6,799 (192 GB/2 TB) and $7,099 (192 GB/4 TB), all variants available, GMKtec US store, read Nov 9, 2026[42]
HP ZGX Nano AI Station (G1n)HP128 GB coherent unified system memory[44]273 GB/s (128 GB LPDDR5x)[45]GB10: up to 1 PFLOP FP4, theoretical, using the sparsity feature; HP and reviewers list 1,000 TOPS FP4 (sparse)[31]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)[46]$7,399.00 at the HP Store for the 4 TB model; $4,759 on Amazon (Notebookcheck, review updated Oct 5, 2026)[47]
Lenovo ThinkStation PGXLenovo128 GB LPDDR5X unified (256-bit)[48]273 GB/s[48]——US from $5,079 (1 TB base model) on Lenovo's site, per TechRadar's review (Jun 9, 2026); UK from GBP 5,200[49]
MSI EdgeMesa N AI+ (128 GB, X-005US)MSI128 GB LPDDR5X unified system memory[50]273 GB/s[50]——$7,999.00 (128 GB, 4 TB, X-005US, MSI US store, read Nov 11, 2026)[50]
MSI EdgeXpert (MS-C931)MSI64 GB or 128 GB LPDDR5x unified (256-bit)[51]273 GB/s[51]1 PFLOPS FP4 (sparse)[51]—No list price; MSI sells via quote (Get a quote)[51]
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[31]273 GB/s[31]Up to 1 PFLOP FP4 (theoretical, using the sparsity feature) (sparse)[31]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.[52]$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[53]
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)[54]273 GB/s (256-bit; NVIDIA spec table covers both 64 GB and 128 GB)[31]1 PFLOPS FP4 (sparse; NVIDIA spec table covers both memory configurations)[31]—from $4,999 (partner systems, Oct 2026; no Founders Edition)[54]
Sipeed MaixStationSipeed128 GB LPDDR5X (256-bit)[28]273 GB/s (Sipeed claim)[28]256 TOPS FP4 (native FP4, no sparsity; Axera/Sipeed claim)[28]—about $3,999 expected for 128 GB, with Kickstarter launch discounts (Sipeed to Tom's Hardware, Oct 2026)[28]
AMD Ryzen AI Halo Developer PlatformAMD128 GB LPDDR5x-8000 unified[55]256 GB/s[55]—gpt-oss-120b, vLLM online serving, equal ISL/OSL 256/256: 222 tok/s aggregate at batch 64 vs 701 on DGX Spark (StorageReview)[56]$3,999 with 2 TB SSD (Jul 2026)[56]
Framework Desktop (Ryzen AI Max+ 395)Framework128 GB LPDDR5x-8000 soldered (32 GB and 64 GB options)[57]256 GB/s[58]—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)[59]$3,449 (DIY Edition, Max+ 395 with 128 GB, system only; out of stock, Nov 3, 2026)[60]
Acer Veriton GN100 AI Mini WorkstationAcer128 GB LPDDR5x coherent unified[61]——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)[62]from $3,999 in North America (announced Sep 3, 2025; prices vary by region)[61]
ASRock Industrial AI BOX-A395ASRock Industrial128 GB LPDDR5X-8000 soldered[63]——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[64]—
Beelink GTR9 Pro (Ryzen AI Max+ 495, 128 GB)Beelink128 GB LPDDR5X-8533 unified (soldered)[65]—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[65]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)[65]$4,799 (128 GB/2 TB), Beelink store, pre-order (read Nov 13, 2026)[65]
Beelink GTR9 Pro (Ryzen AI Max+ 495, 192 GB)Beelink192 GB LPDDR5X-8533 unified (soldered)[65]—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[65]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)[65]$7,399 (192 GB/2 TB) or $7,799 (192 GB/4 TB), Beelink store, pre-order (read Nov 13, 2026)[65]
Bosgame M5 MAX AI Mini PCBosgame192 GB unified LPDDR5 (soldered); up to 160 GB allocatable to the GPU[66]———$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[66]
Dell Pro Max Micro (FCM2250)Dell24 GB GDDR7 on the RTX PRO 4000 SFF (23.9 GB usable measured) + 96 GB DDR5-6400 CSODIMM system RAM in StorageReview's tested build[67]——Ministral 3 8B, llama.cpp server (CUDA), 4,096-in/1,024-out: 52.0 tok/s at 1 request, 108.9 at 4; TTFT 1,477 ms (StorageReview, RTX PRO 4000 SFF 24 GB, driver 595.99.02, Ubuntu 26.04.1)[67]From $1,449.00 on Dell.com US; RTX PRO 4000 SFF 24 GB option +$3,870.00 (read Nov 16, 2026)[68]
Dell Pro Precision 9 T4DellUp to 2x 96 GB GDDR7 (two RTX PRO 6000 Max-Q) + up to 4 TB DDR5 RDIMM in 16 slots, 8 channels[69]——MiniMax M2.7 Q4_K_M (129.3 GB) split across 2x RTX PRO 6000 Max-Q: 83.7 tok/s one user, 127.3 peak; 30.3 on one card plus system RAM (StorageReview, llama.cpp b11716, Ubuntu 26.04.1)[69]From $5,925.00 on Dell.com US; dual RTX PRO 6000 Max-Q option +$30,820.00 (read Nov 14, 2026). StorageReview's tested build (Xeon 698X, 256 GB, 2x Max-Q) about $66,600[70]
Dell Pro Precision with GB300 (model FCT6263; announced as Dell Pro Max with GB300)Dell252 GB HBM3e GPU + 496 GB LPDDR5X CPU[71]———$160,195.00 (Dell.com US default config FCT6263, read Nov 8, 2026)[71]
Gigabyte AI TOP Aeroup to 128 GB LPDDR5X unified[72]————
GMKtec EVO-X2 AI Mini PCGMKtec64 GB or 128 GB onboard LPDDR5X-8000 (non-upgradeable)[73]——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)[74]$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)[73]
HP Z2 Mini G1a WorkstationHPup to 128 GB unified, up to 96 GB assignable to the GPU, up to 8000 MT/s[75]—up to 50 NPU TOPS[75]gpt-oss-120b GGUF (63.39 GB, full GPU offload) in LM Studio: close to 40 tok/s (StorageReview)[76]$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)[76]
Microsoft Surface RTX Spark Dev BoxMicrosoft128 GB unified memory[77]—up to 1 PFLOP of AI compute (Microsoft; precision and sparsity not stated on the page) (sparse)[77]—From $5,999.99 (Microsoft Store US, pre-order, read Nov 1, 2026)[77]
MSI EdgeMesa N AI+ (64 GB)————from $4,699 in the US (64 GB soldered LPDDR5X, 1 TB PCIe Gen 5)[78]
Tenstorrent TT-QuietBox 2 (Blackhole)Tenstorrent128 GB GDDR6 on the accelerators + 256 GB DDR5-5600 system memory[79]———$9,999 (Oct 30, 2026)[79]
WiCi One (RTX 5060 Ti edition)WiCi16 GB GDDR7 VRAM; 1 TB NVMe 4 SSD for TurboStream weight paging (WiCi's claim: NVMe used as working VRAM). System RAM not stated[80]———$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)[80]
WiCi One (RTX 5090 edition)WiCi32 GB GDDR7 VRAM; 4 TB NVMe 5 SSD for TurboStream weight paging (WiCi's claim: NVMe used as working VRAM). System RAM not stated[80]———$9,999 list (WiCi shop, Oct 2026); a $600-off code (WICIWELCOME) is applied at checkout[80]

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 (80)
  1. [1] ASUS
  2. [2] ASUS
  3. [3] StorageReview
  4. [4] Central Computers
  5. [5] HP
  6. [6] StorageReview
  7. [7] HP
  8. [8] StorageReview
  9. [9] Newegg
  10. [10] NVIDIA
  11. [11] Exxact
  12. [12] Supermicro
  13. [13] ServeTheHome
  14. [14] ServeTheHome
  15. [15] TechRadar
  16. [16] NVIDIA
  17. [17] StorageReview
  18. [18] NVIDIA
  19. [19] Apple
  20. [20] Tom's Hardware
  21. [21] MacRumors
  22. [22] Corsair
  23. [23] StorageReview
  24. [24] NVIDIA
  25. [25] ServeTheHome
  26. [26] StorageReview
  27. [27] NVIDIA
  28. [28] Tom's Hardware
  29. [29] ASUS
  30. [30] StorageReview
  31. [31] NVIDIA
  32. [32] ASUS eShop
  33. [33] Dell
  34. [34] HotHardware
  35. [35] StorageReview
  36. [36] Dell
  37. [37] Framework Blog
  38. [38] Framework
  39. [39] GIGABYTE (PR Newswire)
  40. [40] StorageReview
  41. [41] Newegg
  42. [42] GMKtec
  43. [43] GMKtec
  44. [44] HP
  45. [45] HotHardware
  46. [46] StorageReview
  47. [47] Notebookcheck
  48. [48] Lenovo PSREF
  49. [49] TechRadar
  50. [50] MSI
  51. [51] MSI
  52. [52] Tom's Hardware
  53. [53] NVIDIA Marketplace
  54. [54] NVIDIA Blog
  55. [55] AMD
  56. [56] StorageReview
  57. [57] Framework
  58. [58] Framework Blog
  59. [59] ServeTheHome
  60. [60] Framework
  61. [61] Acer
  62. [62] StorageReview
  63. [63] ServeTheHome
  64. [64] ServeTheHome
  65. [65] Beelink
  66. [66] Notebookcheck
  67. [67] StorageReview
  68. [68] Dell
  69. [69] StorageReview
  70. [70] Dell
  71. [71] Dell
  72. [72] Notebookcheck
  73. [73] GMKtec
  74. [74] ServeTheHome
  75. [75] HP
  76. [76] StorageReview
  77. [77] Microsoft
  78. [78] Notebookcheck
  79. [79] Tenstorrent
  80. [80] WiCi

Sources

Published
Sources
2 cited