NVIDIA GeForce RTX 4090 48 GB
Live eBay prices, cheapest first. The best one right now is $5,469 (new).
Other sizes: 24 GB
Specs
- VRAM
- 48 GB GDDR6X
- Bandwidth
- 1,008 GB/s
- FP32
- 82.6 TFLOPS
- FP16
- 165 TFLOPS
- Power
- 450 W
- Connector
- PCIe 4.0
- Released
- 2022
- Launch price
- $1,599
- CUDA compute
- 8.9
- Llama 3.1 8B at Q4 (est.)
- 104–156 tokens/s
Price right now
Live on eBayFrom the live listings below; shipping included.
Listings
Live on eBay| Listing | Condition | Price | Checked |
|---|---|---|---|
| New | $5,469−12%shipping not listed | ||
| New | $5,699−9%free shipping | ||
| New | $6,250or best offerfree shipping | ||
| New | $6,450or best offerfree shipping | ||
| New | $6,500or 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- #12 of 19 PCIe GPUs with 48 GB or more for the money: $3,457 per GeForce RTX 3090 of performance.
- Software: CUDA 11.8 and newer.
As much VRAM, about as fast, for less
- AMD Instinct MI210: $3,799, 211% of the GeForce RTX 3090, 64 GB
- NVIDIA GeForce RTX 4090 D 48 GB: $5,000, 149% of the GeForce RTX 3090, 48 GB
Faster, for the least
- AMD Instinct MI100: $1,240, 185% of the GeForce RTX 3090, 32 GB
- 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
More VRAM, for not much more
- NVIDIA A16: $2,200, 93% of the GeForce RTX 3090, 64 GB
- AMD Instinct MI210: $3,799, 211% of the GeForce RTX 3090, 64 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 4090 48 GB needed | Costs now | Speed (est.) |
|---|---|---|---|
| Llama 3.1 8B | 1 card | $5,469 | 104–156 tokens/s |
| Qwen3 14B | 1 card | $5,469 | 56–84 tokens/s |
| gpt-oss 20B | 1 card | $5,469 | 139–208 tokens/s |
| Gemma 3 27B | 1 card | $5,469 | 30–46 tokens/s |
| Qwen3 32B | 1 card | $5,469 | 25–38 tokens/s |
| Llama 3.3 70B | 1 card | $5,469 | 12–18 tokens/s |
| gpt-oss 120B | 2 cards | $11,168 | 98–147 tokens/s |
| Qwen3 235B-A22B | 4 cards | $23,868 | 23–34 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 4090 48 GB needed | Costs now | Per question (est.) |
|---|---|---|---|
| Julia-10.144B | 1 card | $5,469 | 4.9–10 ms |
| Laya0.421B | 1 card | $5,469 | 8.1–17 ms |
| Kev-4B4B | 1 card | $5,469 | 19–42 ms |
| lev4B | 1 card | $5,469 | 58–125 ms |
| OpenJev27B | 1 card | $5,469 | 70–150 ms |
| Clef27B | 1 card | $5,469 | 70–150 ms |
| Perplexity Decider v1.227B | 1 card | $5,469 | 70–150 ms |
| mammoth-127B | 1 card | $5,469 | 70–150 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 4090 48 GB. These open the shop's own search; we don't show their prices here.




