PerGB

AMD Instinct MI50 16 GB

Live eBay prices, cheapest first. The best one right now is $266.00 (used).

Other sizes: 32 GB

Specs

VRAM
16 GB HBM2
Bandwidth
1,024 GB/s
FP32
13.4 TFLOPS
FP16
26.8 TFLOPS
Power
300 W
Connector
PCIe 4.0
Released
2018
Llama 3.1 8B at Q4 (est.)
74–127 tokens/s

gfx906: dropped from current ROCm; runs llama.cpp (ROCm/Vulkan) and older ROCm builds.

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
Two 8-pin PCIe power inputs. Rated 300 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.

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$266.00
Median ask$309.30
Per GB VRAM$16.62
vs launch price—
Live listings21

From the live listings below; shipping included.

Listings

Live on eBay
ListingConditionPriceChecked
Used$266.00−14%free shipping
Used$272.62or best offer−12%+$60.00 shipping
Used$273.00−12%+$5.00 shipping
Used$275.00−11%shipping not listed
Used$297.60free shipping
Used$300.00or best offershipping not listed
Used$302.78or best offer+$84.00 shipping
Used$305.90+$38.00 shipping
Used$306.98+$40.00 shipping
Used$308.99or best offer+$9.99 shipping
Used$309.30+$40.00 shipping
Used$309.30+$40.00 shipping
Used$309.30+$40.00 shipping
Used$309.30+$40.00 shipping
Used$309.30+$40.00 shipping
Used$334.50+$155.00 shipping
Used$334.50+$155.00 shipping
Used$339.98+$40.00 shipping
Used$344.50+$165.00 shipping
Used$420.60+$30.00 shipping
Used$730.00or best offer+$300.00 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

As much VRAM, about as fast, for less

Faster, for the least

More VRAM, for not much more

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)Instinct MI50 16 GB neededCosts nowSpeed (est.)
Llama 3.1 8B1 card$266.0074–127 tokens/s
Qwen3 14B1 card$266.0040–68 tokens/s
gpt-oss 20B1 card$266.0099–169 tokens/s
Gemma 3 27B2 cards$538.6222–37 tokens/s
Qwen3 32B2 cards$538.6218–31 tokens/s
Llama 3.3 70B4 cards$1,0878.4–14 tokens/s
gpt-oss 120B5 cards$1,38470–119 tokens/s
Qwen3 235B-A22Bmore than 8——
DeepSeek R1 / V3 671Bmore than 8——

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)Instinct MI50 16 GB neededCosts nowPer question (est.)
Julia-10.144B1 card$266.0016–34 ms
Laya0.421B1 card$266.0026–56 ms
Kev-4B4B1 card$266.0063–135 ms
lev4B1 card$266.00189–406 ms
OpenJev27B2 cards$538.62226–485 ms
Clef27B2 cards$538.62226–485 ms
Perplexity Decider v1.227B2 cards$538.62226–485 ms
mammoth-127B2 cards$538.62226–485 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 AMD Instinct MI50 16 GB. These open the shop's own search; we don't show their prices here.