NVIDIA GeForce RTX 5090 24 GB
Live eBay prices, cheapest first. The best one right now is $6,490 (open box).
Other sizes: 32 GB
Specs
- VRAM
- 24 GB GDDR7
- Bandwidth
- 1,792 GB/s
- FP32
- 104.8 TFLOPS
- FP16
- 209 TFLOPS
- Power
- 575 W
- Connector
- PCIe 5.0
- Released
- 2025
- Launch price
- $1,999
- CUDA compute
- 12.0
- Llama 3.1 8B at Q4 (est.)
- 185–277 tokens/s
Price right now
Live on eBayFrom the live listings below; shipping included.
Listings
Live on eBay| Listing | Condition | Price | Checked |
|---|---|---|---|
| Open box | $6,490or best offerfree shipping | ||
| Open box | $6,490or 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- Software: CUDA 12.8 and newer.
As much VRAM, about as fast, for less
- NVIDIA GeForce RTX 5090: $5,000, 237% of the GeForce RTX 3090, 32 GB
- NVIDIA A100 PCIe 40 GB: $5,588, 270% of the GeForce RTX 3090, 40 GB
Faster, for the least
- NVIDIA A100 PCIe 80 GB: $10,999, 301% of the GeForce RTX 3090, 80 GB
- NVIDIA RTX PRO 6000 Blackwell: $13,100, 367% of the GeForce RTX 3090, 96 GB
- NVIDIA H100 PCIe: $26,935, 477% of the GeForce RTX 3090, 80 GB
More VRAM, for not much more
- AMD Instinct MI50 32 GB: $633.68, 64% of the GeForce RTX 3090, 32 GB
- AMD Radeon Pro W6800: $879.99, 52% of the GeForce RTX 3090, 32 GB
- NVIDIA Tesla V100 PCIe 32 GB: $975.00, 123% of the GeForce RTX 3090, 32 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) | GeForce RTX 5090 24 GB needed | Costs now | Speed (est.) |
|---|---|---|---|
| Llama 3.1 8B | 1 card | $6,490 | 185–277 tokens/s |
| Qwen3 14B | 1 card | $6,490 | 100–150 tokens/s |
| gpt-oss 20B | 1 card | $6,490 | 246–369 tokens/s |
| Gemma 3 27B | 1 card | $6,490 | 54–81 tokens/s |
| Qwen3 32B | 1 card | $6,490 | 45–68 tokens/s |
| Llama 3.3 70B | 2 cards | $12,980 | 21–31 tokens/s |
| gpt-oss 120B | 4 cards | not enough listings | 174–261 tokens/s |
| Qwen3 235B-A22B | 7 cards | not enough listings | 40–60 tokens/s |
| DeepSeek R1 / V3 671B | more 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) | GeForce RTX 5090 24 GB needed | Costs now | Per question (est.) |
|---|---|---|---|
| Julia-10.144B | 1 card | $6,490 | 3.2–7.0 ms |
| Laya0.421B | 1 card | $6,490 | 5.4–12 ms |
| Kev-4B4B | 1 card | $6,490 | 13–28 ms |
| lev4B | 1 card | $6,490 | 39–84 ms |
| OpenJev27B | 1 card | $6,490 | 47–100 ms |
| Clef27B | 1 card | $6,490 | 47–100 ms |
| Perplexity Decider v1.227B | 1 card | $6,490 | 47–100 ms |
| mammoth-127B | 1 card | $6,490 | 47–100 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 GeForce RTX 5090 24 GB. These open the shop's own search; we don't show their prices here.

