AMD Instinct MI210
Live eBay prices, cheapest first. The best one right now is $3,799 (used).
Specs
- VRAM
- 64 GB HBM2e
- Bandwidth
- 1,638 GB/s
- FP32
- 22.6 TFLOPS
- FP16
- 181 TFLOPS
- Power
- 300 W
- Connector
- PCIe 4.0
- Released
- 2022
- Launch price
- $10,000 (est.)
- Llama 3.1 8B at Q4 (est.)
- 118–203 tokens/s
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.
- 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 eBayFrom the live listings below; shipping included.
Listings
Live on eBay| Listing | Condition | Price | Checked |
|---|---|---|---|
| Used | $3,799−12%free shipping | ||
| Used | $3,799−12%free shipping | ||
| Used | $3,800−12%free shipping | ||
| Used | $3,851−10%free shipping | ||
| Used | $3,885−10%free shipping | ||
| Used | $3,969or best offer−8%free shipping | ||
| Used | $3,969or best offer−8%free shipping | ||
| Used | $3,985or best offer−7%free shipping | ||
| Used | $3,985or best offer−7%free shipping | ||
| Used | $3,985or best offer−7%free shipping | ||
| Used | $3,985or best offer−7%free shipping | ||
| Used | $4,050−6%free shipping | ||
| Open box | $4,099free shipping | ||
| Open box | $4,100or best offerfree shipping | ||
| Used | $4,187or best offerfree shipping | ||
| Used | $4,220+$20.00 shipping | ||
| Used | $4,299free shipping | ||
| Used | $4,369or best offerfree shipping | ||
| Used | $4,400or best offerfree shipping | ||
| Open box | $4,617or best offerfree shipping | ||
| Open box | $4,920+$20.00 shipping | ||
| Used | $5,039or best offerfree shipping | ||
| Used | $5,043or best offerfree shipping | ||
| Used | $5,044or best offerfree shipping | ||
| Used | $5,054or best offerfree shipping | ||
| Used | $5,232+$20.00 shipping | ||
| Open box | $5,450free shipping | ||
| Used | $5,800or best offerfree shipping | ||
| Used | $5,950shipping not listed | ||
| New | $6,000or best offerfree shipping | ||
| New | $6,820or best offer+$120.00 shipping | ||
| Open box | $7,208+$20.00 shipping | ||
| Used | $8,729or 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- #1 of 8 PCIe GPUs with 64 GB or more for the money: $1,799 per GeForce RTX 3090 of performance.
- Software: ROCm (on AMD's support list).
Faster, for the least
- NVIDIA A100 PCIe 40 GB: $5,588, 270% of the GeForce RTX 3090, 40 GB
- NVIDIA RTX PRO 5000 Blackwell 48 GB: $6,590, 243% of the GeForce RTX 3090, 48 GB
- NVIDIA RTX PRO 5000 Blackwell 72 GB: $8,999, 243% of the GeForce RTX 3090, 72 GB
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 MI210 needed | Costs now | Speed (est.) |
|---|---|---|---|
| Llama 3.1 8B | 1 card | $3,799 | 118–203 tokens/s |
| Qwen3 14B | 1 card | $3,799 | 64–110 tokens/s |
| gpt-oss 20B | 1 card | $3,799 | 158–270 tokens/s |
| Gemma 3 27B | 1 card | $3,799 | 35–59 tokens/s |
| Qwen3 32B | 1 card | $3,799 | 29–49 tokens/s |
| Llama 3.3 70B | 1 card | $3,799 | 13–23 tokens/s |
| gpt-oss 120B | 2 cards | $7,598 | 111–191 tokens/s |
| Qwen3 235B-A22B | 3 cards | $11,398 | 26–44 tokens/s |
| DeepSeek R1 / V3 671B | 7 cards | $27,072 | 15–26 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 MI210 needed | Costs now | Per question (est.) |
|---|---|---|---|
| Julia-10.144B | 1 card | $3,799 | 4.8–10 ms |
| Laya0.421B | 1 card | $3,799 | 8.0–17 ms |
| Kev-4B4B | 1 card | $3,799 | 19–41 ms |
| lev4B | 1 card | $3,799 | 58–124 ms |
| OpenJev27B | 1 card | $3,799 | 69–148 ms |
| Clef27B | 1 card | $3,799 | 69–148 ms |
| Perplexity Decider v1.227B | 1 card | $3,799 | 69–148 ms |
| mammoth-127B | 1 card | $3,799 | 69–148 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 MI210. These open the shop's own search; we don't show their prices here.




























