The NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition costs about $15,000 to $18,300 in the US as of 4 October 2026, when you can find it. Dell lists it at $18,272.99 in limited stock. Newegg lists $17,999.99. PNY’s retail box was $14,999 on a tracker, but out of stock.
It is a 300 W card with 96 GB of GDDR7 ECC memory and up to four cards per PC. It is the same chip as the 600 W Workstation Edition, run at half the power. Puget Systems measured it 5 to 14 percent slower.
In a hurry? Jump to: price · what Max-Q means · vs Workstation Edition · how many cards fit · power supply · specs · who should buy · how we checked
How much does the RTX PRO 6000 Max-Q cost? (updated October 13, 2026)
October 13 update: US prices have fallen since our first check. The rtx.live tracker shows the NVIDIA-branded single-fan OEM card (900-5G153-2200-000) at $14,800 at Micro Center, the lowest price it has recorded for a new card sold by a store itself. PNY’s single- and dual-fan retail boxes start at $16,000. Over the tracker’s first 16 days, starting September 28, the daily low ranged from $14,800 to $15,900. The October 4 figures below are kept for comparison.
NVIDIA’s Max-Q page shows no price and sends you to partners. Here is what we could see for single cards:
| Where | Price | Notes |
|---|---|---|
| Dell (PNY VCNRTXPRO6000BQ-PB) | $18,272.99 | Limited stock. We opened the page on 4 October. |
| Newegg (NVIDIA box, 900-5G153-2200-000) | $17,999.99 | Available, limit one per customer. Opened 4 October. |
| NeweggBusiness (same part) | $17,999.99 | In stock. Opened 4 October. |
| Newegg / PNY retail box (VCNRTXPRO6000BQ-PB) | $14,999.00 | Out of stock on 29 September, per a stock tracker. |
| B&H Photo (900-5G153-2200-000) | $17,499 | Out of stock. The tracker last saw it on 27 May, so treat it as old. |
| Dell Pro Precision 9 T6 (configurator) | +$24,530 for two; +$36,890 for three | Upgrade prices over the base GPU. Dell US site, 4 October. |
A stock tracker also lists NeweggBusiness at about $8,290. We could not confirm that. The live NeweggBusiness page showed $17,999.99. It may be an old promo-code price. Do not plan around it.
Why so high? The sister card, the Workstation Edition, went from $8,565 at launch to $16,000 on NVIDIA’s store, per Tom’s Hardware. Tom’s ties the rises to memory prices. We found no NVIDIA Max-Q price history. Our report on rising GPU prices covers the wider squeeze.
Buying tips:
- Match the part number: 900-5G153-2200-000 (NVIDIA box) or VCNRTXPRO6000BQ-PB (PNY box). 300 W means Max-Q. 600 W is the other card.
- The PNY box has a 3-year limited warranty.
- Stock comes and goes. Set alerts instead of trusting last month’s price.
- For two to four cards, compare a system maker’s price with loose cards.
If 32 GB is enough, other pro cards cost far less. Intel’s 32 GB Arc Pro B70 is one, though its price has climbed too.
What does Max-Q mean on the RTX PRO 6000?
It is not a cut-down chip. The Max-Q has the same 24,064 CUDA cores, 96 GB of GDDR7 ECC, 512-bit bus and 1,792 GB/s of bandwidth as the Workstation Edition. Three things change:
- Power: 300 W instead of 600 W.
- Cooler: an active cooler that Puget describes as blower-style, like the RTX 6000 Ada. It pulls air from the case and blows it out the back.
- Size: 10.5 in long and full height, instead of 12 in long and extended height.
The rear exhaust is why you can put several cards side by side. Heat leaves the case instead of warming the next card.
RTX PRO 6000 Max-Q vs Workstation Edition
NVIDIA’s Max-Q datasheet and its Workstation datasheet differ in these rows:
| Max-Q | Workstation Edition | |
|---|---|---|
| Board power | 300 W | 600 W |
| AI TOPS (FP4 sparse) | 3,511 | 4,000 |
| FP32 | 110 TFLOPS | 126 TFLOPS |
| RT performance | 333 TFLOPS | 382 TFLOPS |
| Cooling | Active (blower-style) | Double flow-through |
| Size | 4.4 x 10.5 in, full height | 5.4 x 12 in, extended height |
| Memory and bandwidth | 96 GB, 1,792 GB/s | 96 GB, 1,792 GB/s |
| Cards per PC | Up to 4 (NVIDIA) | 1 or 2 (Dell sells dual) |
Puget Systems tested both cards on one PC in July 2025. The Max-Q was 5 to 13 percent slower in Blender, V-Ray, Octane and Redshift, 8% in DaVinci Resolve overall, 5 to 9 percent in After Effects, 13% in Topaz Video AI and 14% in Unreal Engine.
Our arithmetic on NVIDIA’s numbers: the Max-Q gives about 11.7 AI TOPS per watt against 6.7 for the Workstation Edition. That is roughly 1.75 times more per watt. It is not a benchmark.
Which one? Pick the Max-Q for two to four cards or a smaller case. Pick the Workstation Edition for one fast card. For racks, read the Server Edition guide.
How many RTX PRO 6000 Max-Q cards fit in one PC?
NVIDIA says you can scale from one to four cards, for up to 384 GB of combined memory. Dell’s Pro Precision 9 T6 lets you pick dual and triple Max-Q setups. Puget advises one empty slot between cards when the board allows it.
Two cautions. First, 384 GB is four separate 96 GB pools, not one. The cards have no NVLink; Lenovo’s guide says NVLink support is No. They talk over PCIe 5.0, and your software must split the model.
Second, not every app scales. In Puget’s Resolve test, three Max-Q cards gave about 2x on GPU effects. But AI features barely scaled, and one test ran slower than a single card because of card conflicts.
What power supply and circuit does it need?
One Max-Q draws 300 W through a single 16-pin connector. NVIDIA gives no power supply size. These are our planning numbers, not NVIDIA’s. They leave room for a Threadripper or Xeon, drives and fans:
| Cards | GPU power | Our PSU rule of thumb |
|---|---|---|
| 1 | 300 W | 850 W or more |
| 2 | 600 W | 1,200 W or more |
| 3 | 900 W | 1,600 W or more |
| 4 | 1,200 W | 2,000 W or more, or two supplies |
A US 15 A, 120 V outlet gives 1,800 W at most, or about 1,440 W for steady loads. Three or four cards can pass that. Plan a 20 A or 240 V circuit.
Noise: we found no published noise test for the Max-Q. GamersNexus measured the Workstation Edition at about 32.5 dBA. Treat the Max-Q’s noise as unknown until you hear one.
RTX PRO 6000 Max-Q specs
From NVIDIA’s datasheet (April 2025), PNY’s page and Lenovo’s product guide (updated 8 May 2026).
| Spec | Max-Q Workstation Edition |
|---|---|
| GPU architecture | NVIDIA Blackwell, TSMC 4N, 92.2 billion transistors (PNY) |
| CUDA cores | 24,064 |
| Tensor / RT cores | 752 5th-gen / 188 4th-gen (PNY) |
| AI performance | 3,511 AI TOPS (FP4 with sparsity) |
| FP32 / RT | 110 TFLOPS / 333 TFLOPS |
| Memory | 96 GB GDDR7 with ECC, 512-bit bus |
| Memory bandwidth | 1,792 GB/s |
| Interface | PCIe 5.0 x16 |
| NVLink | No (Lenovo) |
| Display outputs | 4x DisplayPort 2.1b |
| Max displays | 4x 4096 x 2160 at 120 Hz · 4x 5120 x 2880 at 60 Hz · 2x 7680 x 4320 at 60 Hz |
| Video engines | 4x NVENC (9th gen), 4x NVDEC (6th gen) |
| MIG | Up to 4x 24 GB, 2x 48 GB or 1x 96 GB |
| vGPU | No support (Lenovo) |
| Total board power | 300 W |
| Power connector | 1x 16-pin (PCIe CEM5) |
| Cooling | Active (blower-style per Puget and PNY) |
| Size | 4.4 in high x 10.5 in long, dual slot, full height |
| Part numbers | NVIDIA 900-5G153-2200-000 · PNY VCNRTXPRO6000BQ-PB |
Lenovo’s guide lists the DisplayPorts as off by default on its version. If a new card gives no picture, check NVIDIA’s Display Mode Selector tool. NVIDIA warns that misuse can permanently damage the GPU.
Who should buy the Max-Q, and who should skip it?
Buy it if:
- You want two to four cards in one workstation.
- Your case or power supply cannot take a 600 W, 12 inch card.
- You want 96 GB of ECC memory and can live with 5 to 14 percent less speed.
Skip it if:
- You want one card and the best speed. Buy the Workstation Edition.
- Your work fits in 32 GB. An RTX 5090 costs far less.
- You need vGPU. The Max-Q does not support it.
- You are building a rack. The Server Edition is made for that.
Where these numbers come from
- We read NVIDIA’s Max-Q datasheet (April 2025) and product page, PNY’s page, and Lenovo’s guide (updated 8 May 2026).
- On 4 October 2026 we opened the Dell, Newegg and NeweggBusiness listings and the Dell configurator, and read stock trackers.
- Speed differences are Puget Systems’ July 2025 numbers. We cite them; we did not rerun them.
- The PSU and circuit figures are our own arithmetic. They are not from NVIDIA.
- We did not test this card. There is no hands-on testing and no benchmark of our own here.
- Not confirmed: any NVIDIA Max-Q price, the NeweggBusiness price near $8,290, noise levels, and the maximum card count in each Dell model.
Should you buy one?
The RTX PRO 6000 Max-Q is the same 96 GB chip at half the power. It is the right card when you want several in one PC. At about $15,000 to $18,300 and often short on stock, shop carefully and match the part number.
Only need one card? Read the Workstation Edition guide. Planning a rack? Read the Server Edition guide.
Frequently Asked Questions
How much does the RTX PRO 6000 Blackwell Max-Q cost?
About $15,000 to $18,300 in the US on 4 October 2026, when in stock. Dell listed $18,272.99, Newegg $17,999.99, and PNY's retail box $14,999 (out of stock on 29 September). NVIDIA lists no Max-Q price. A tracker shows about $8,290 at NeweggBusiness, but we could not confirm it.
What does Max-Q mean on the RTX PRO 6000?
It is the 300 W version of the same chip. You get the same 24,064 CUDA cores, 96 GB of GDDR7 ECC and 1,792 GB/s as the 600 W Workstation Edition. It has a blower-style cooler and is 10.5 inches long, built for putting several cards in one PC.
Max-Q vs Workstation Edition: how much slower is it?
Puget Systems measured the Max-Q 5 to 13 percent slower in Blender, V-Ray, Octane and Redshift, 8% in DaVinci Resolve overall, 13% in Topaz Video AI and 14% in Unreal Engine. NVIDIA rates it at 3,511 AI TOPS against 4,000, on half the power.
How many RTX PRO 6000 Max-Q cards fit in one workstation?
NVIDIA says up to four. Dell's Pro Precision 9 T6 configurator offers dual and triple setups. Puget suggests leaving one empty slot between cards. Four cards total 384 GB, but as four separate 96 GB pools because there is no NVLink.
What power supply does the RTX PRO 6000 Max-Q need?
NVIDIA gives no size. One card draws 300 W through a 16-pin connector. Our rule of thumb: 850 W for one card, 1,200 W for two, 1,600 W for three and 2,000 W or more for four. A 15 A, 120 V US outlet tops out near 1,800 W.
Will the RTX PRO 6000 Max-Q fit in a small case?
It is 10.5 inches long, 4.4 inches tall and dual slot, so it fits more cases than the 12 inch Workstation Edition. It is not a small-form-factor card. Leave room behind it for the 16-pin cable bend.
Does the RTX PRO 6000 Max-Q support NVLink or vGPU?
No to both. Lenovo's guide lists NVLink support as No and vGPU as no support. MIG is supported, with up to four 24 GB instances. For vGPU, look at the Server Edition instead.
Is the RTX PRO 6000 Max-Q worth it over an RTX 5090?
Yes if you need more than 32 GB, ECC memory, pro drivers or several cards in one PC. The Max-Q has 96 GB against 32 GB. If your work fits in 32 GB, the 5090 costs far less.
