PerGB

AMD Instinct MI250X

Live eBay prices, cheapest first. The best one right now is $3,632 (used).

Not a PCIe card. OAM module: needs an OAM baseboard/server; not practical in a regular motherboard. No OAM-to-PCIe adapter runs it yet (November 2026): hobbyists get MI250 / MI250X modules to power up on one, but not past the driver's start-up.

Specs

VRAM
128 GB HBM2e
Bandwidth
3,277 GB/s
FP32
47.9 TFLOPS
FP16
383 TFLOPS
Power
560 W
Connector
OAM (PCIe 4.0)
Released
2021
Llama 3.1 8B at Q4 (est.)
236–405 tokens/s

Price right now

Live on eBay
Lowest now$3,632
Median ask$4,816
Per GB VRAM$28.38
vs launch price—
Live listings2

From the live listings below; shipping included.

Listings

Live on eBay
ListingConditionPriceChecked
Used$3,632or best offerfree shipping
Used$6,000or 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

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 MI250X neededCosts nowSpeed (est.)
Llama 3.1 8B1 card$3,632236–405 tokens/s
Qwen3 14B1 card$3,632128–219 tokens/s
gpt-oss 20B1 card$3,632315–541 tokens/s
Gemma 3 27B1 card$3,63269–118 tokens/s
Qwen3 32B1 card$3,63258–99 tokens/s
Llama 3.3 70B1 card$3,63227–46 tokens/s
gpt-oss 120B1 card$3,632223–382 tokens/s
Qwen3 235B-A22B2 cards$9,63252–88 tokens/s
DeepSeek R1 / V3 671B4 cardsnot enough listings31–53 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)Instinct MI250X neededCosts nowPer question (est.)
Julia-10.144B1 card$3,6322.3–5.0 ms
Laya0.421B1 card$3,6323.9–8.3 ms
Kev-4B4B1 card$3,6329.3–20 ms
lev4B1 card$3,63228–60 ms
OpenJev27B1 card$3,63233–72 ms
Clef27B1 card$3,63233–72 ms
Perplexity Decider v1.227B1 card$3,63233–72 ms
mammoth-127B1 card$3,63233–72 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 MI250X. These open the shop's own search; we don't show their prices here.