PerGB

NVIDIA A16

Live eBay prices, cheapest first. The best one right now is $2,200 (used).

Specs

VRAM
64 GB GDDR6
Bandwidth
800 GB/s
FP32
18 TFLOPS
FP16
72 TFLOPS
Power
250 W
Connector
PCIe 4.0
Released
2021
Launch price
$3,000 (est.)
CUDA compute
8.6
Llama 3.1 8B at Q4 (est.)
82–124 tokens/s

Four small GPUs with 16 GB each on one card (VDI card).

Running it at home

Cooling
Passive: it has no fan. In a desktop it needs air pushed through it, front to back: a fan shroud on its end (3D-printed ones with a 40 mm blower or a 92 mm fan are sold for most of these cards) or a server chassis. Without one it overheats within minutes under load.
Power
One 8-pin CPU (EPS-12V) power input, not a PCIe one: it needs an adapter cable (two PCIe 8-pin to one EPS 8-pin, sold for these cards). A PCIe plug fits but is wired differently: never force one in. Rated 250 W: allow that for each card on top of the rest of the PC.
Size
Full height, two slots.
Motherboard
Turn on "Above 4G decoding" in the BIOS: without it many boards won't start with a card with this much memory.
Screen
No display outputs: use the processor's graphics or a small second card for a monitor.

A server card: made for a rack server's airflow and power. With these sorted it works in a normal PC.

Price right now

Live on eBay
Lowest now$2,200
Median ask$4,160
Per GB VRAM$34.37
vs launch price27% under
Live listings54

From the live listings below; shipping included.

Listings

Live on eBay
ListingConditionPriceChecked
Used$2,200free shipping
Used$2,239shipping not listed
Used$2,350−44%free shipping
Used$2,382or best offer−43%+$66.00 shipping
Used$2,499−40%free shipping
Used$2,500or best offer−40%Check carefullyfree shipping
Used$2,500or best offer−40%Check carefullyfree shipping
Used$2,600−37%shipping not listed
Used$2,800−33%free shipping
Used$2,800−33%free shipping
Used$2,859or best offer−31%free shipping
Used$2,899or best offer−30%free shipping
Used$2,950−29%free shipping
Used$3,250or best offer−22%free shipping
Used$3,255or best offer−22%free shipping
Open box$3,299or best offer−21%free shipping
Used$3,420or best offer−18%+$56.06 shipping
Refurbished$3,424or best offer−18%shipping not listed
Used$3,499or best offer−16%free shipping
Used$3,755or best offer−10%+$79.99 shipping
Used$3,770or best offer−9%free shipping
New$4,049or best offerfree shipping
New$4,049or best offerfree shipping
New$4,049or best offerfree shipping
New$4,049or best offerfree shipping
New$4,049or best offerfree shipping
New$4,139or best offerfree shipping
New$4,181or best offerfree shipping
New$4,354or best offerfree shipping
New$4,400or best offer+$50.00 shipping
New$4,425or best offerfree shipping
New$4,425or best offerfree shipping
New$4,425or best offerfree shipping
New$4,425or best offerfree shipping
New$4,425or best offerfree shipping
New$4,425or best offerfree shipping
New$4,425or best offerfree shipping
New$4,425or best offerfree shipping
New$4,450or best offerfree shipping
Open box$4,495free shipping
Open box$4,495free shipping
Open box$4,495free shipping
New$4,500or best offerfree shipping
New$4,500or best offerfree shipping
Open box$4,510or best offerfree shipping
New$4,542or best offerfree shipping
New$4,719or best offerfree shipping
New$4,719or best offerfree shipping
Open box$4,983+$98.10 shipping
Refurbished$4,983+$98.10 shipping
New$5,403or best offerfree shipping
Refurbished$5,499or best offerfree shipping
New$5,537+$52.00 shipping
New$5,986or best offerfree shipping

Auctions ending soon

No auctions ending soon right now.

Prices include shipping and were read from eBay at the time shown. They can change: check the listing before you buy.

Owner reviews

No owner reviews yet. Own one? Tell others how it runs.

From signed-in owners, checked before they're published. PerGB doesn't pay for or edit reviews.

How it compares

Live on eBay

Faster, for the least

Compare: vs GeForce RTX 3080 Ti, vs GeForce RTX 3090, vs GeForce RTX 4070 Ti Super.

From the live listings of every GPU, shipping included; performance against the GeForce RTX 3090 (FP16 or FP32 compute and memory bandwidth). Updated with every build.

Local AI: what it runs

Model (Q4, 8K context)A16 neededCosts nowSpeed (est.)
Llama 3.1 8B1 card$2,20082–124 tokens/s
Qwen3 14B1 card$2,20045–67 tokens/s
gpt-oss 20B1 card$2,200110–165 tokens/s
Gemma 3 27B1 card$2,20024–36 tokens/s
Qwen3 32B1 card$2,20020–30 tokens/s
Llama 3.3 70B1 card$2,2009.3–14 tokens/s
gpt-oss 120B2 cards$4,43978–116 tokens/s
Qwen3 235B-A22B3 cards$6,78918–27 tokens/s
DeepSeek R1 / V3 671B7 cards$16,67011–16 tokens/s

Cost: that many of the cheapest live listings. Speed: an estimate from the memory bandwidth for one person chatting. Try other models, sizes and contexts.

Decision models

Model (Q4, 8K context)A16 neededCosts nowPer question (est.)
Julia-10.144B1 card$2,2008.3–18 ms
Laya0.421B1 card$2,20014–30 ms
Kev-4B4B1 card$2,20033–71 ms
lev4B1 card$2,20099–213 ms
OpenJev27B1 card$2,200119–254 ms
Clef27B1 card$2,200119–254 ms
Perplexity Decider v1.227B1 card$2,200119–254 ms
mammoth-127B1 card$2,200119–254 ms

Decision models score the options you give them instead of writing text, so they're timed per question. The time is scaled from each model's reported speed on an RTX PRO 6000 by this card's compute and memory bandwidth, within about half either way. About the models.

Compare elsewhere

Search other shops for NVIDIA A16. These open the shop's own search; we don't show their prices here.