GeForce RTX 4090D 48GB Price & Used Prices: Which Local LLMs Can It Run?
Price alert
We'll email you when the Xianyu (CN) price reaches your target.
The GeForce RTX 4090D 48GB has 48GB of GDDR6X memory with 1,008 GB/s of bandwidth. The median asking price is about $3,526 on Xianyu. At 4-bit quantization and 8K context it fits a dense model of up to about 79B parameters, at roughly 15 tokens/s.
Xianyu
GeForce RTX 4090D 48GB price history
Xianyu
Buy links may carry affiliate tags; they do not change the price you pay.
Full GeForce RTX 4090D 48GB specifications
- Memory48 GB · GDDR6X
- Bandwidth1,008 GB/s
- FP16 / BF16294.3 TFLOPS
- FP8 / INT8588.5 / 588.5 TOPS
- Bus width
- 384-bit
- Architecture
- Ada Lovelace (AD102)
- Cores
- 14,592
- Tensor cores
- 456
- Ecosystem
- CUDA
- Interface
- PCIe 4.0 x16
- NVLink
- Not supported
- Power
- 425 W
- Power connector
- 16-pin 12VHPWR
- Slots
- 2
- Type
- Consumer
- Launch date
- Dec 28, 2023
What LLMs can the GeForce RTX 4090D 48GB run?
Models up to 50B parameters on a single GeForce RTX 4090D 48GB, versions released since 2026 only
| Model | Size | Runs? | Speed t/s | Max context |
|---|---|---|---|---|
 Gemma 4 E4B8B | 3.8 GB | Runs | 165.8 | 128K |
 Qwen3.5 9B9.7B | 4.1 GB | Runs | 153.6 | 256K |
| Gemma 4 12B12B | 4.7 GB | Runs | 134.2 | 256K |
| Gemma 4 26B A4B25.2B (3.8B active) | 10.5 GB | Runs | 156.6 | 256K |
| Qwen3.6 27B27.8B | 11.8 GB | Runs | 58.5 | 256K |
| Qwen3.8 27B27.8B | 9.8 GB | Runs | 69.5 | 256K |
 GLM 4.7 Flash30B (3B active) | 11.9 GB | Runs | 169.5 | 198K |
| Gemma 4 31B30.7B | 11.8 GB | Runs | 56.4 | 256K |
| Qwen3.6 35B A3B36B (3B active) | 12.3 GB | Runs | 181.9 | 256K |
| Model | Size | Runs? | Speed t/s | Max context |
|---|---|---|---|---|
| Gemma 4 E4B8B | 5 GB | Runs | 130.7 | 128K |
| Qwen3.5 9B9.7B | 5.7 GB | Runs | 115.3 | 256K |
| Gemma 4 12B12B | 7.1 GB | Runs | 93.5 | 256K |
| Gemma 4 26B A4B25.2B (3.8B active) | 16.9 GB | Runs | 130.1 | 256K |
| Qwen3.6 27B27.8B | 16.8 GB | Runs | 42 | 256K |
| Qwen3.8 27B27.8B | 16.5 GB | Runs | 42.7 | 256K |
| GLM 4.7 Flash30B (3B active) | 18.3 GB | Runs | 147.9 | 198K |
| Gemma 4 31B30.7B | 18.3 GB | Runs | 37.7 | 256K |
| Qwen3.6 35B A3B36B (3B active) | 22.1 GB | Runs | 151.5 | 256K |
| Model | Size | Runs? | Speed t/s | Max context |
|---|---|---|---|---|
| Gemma 4 E4B8B | 8.2 GB | Runs | 83.5 | 128K |
| Qwen3.5 9B9.7B | 9.5 GB | Runs | 72.5 | 256K |
| Gemma 4 12B12B | 12.7 GB | Runs | 54.8 | 256K |
| Gemma 4 26B A4B25.2B (3.8B active) | 26.9 GB | Runs | 102.8 | 256K |
| Qwen3.6 27B27.8B | 28.6 GB | Runs | 25.2 | 256K |
| Qwen3.8 27B27.8B | 29 GB | Runs | 24.8 | 256K |
| GLM 4.7 Flash30B (3B active) | 31.8 GB | Runs | 116.5 | 198K |
| Gemma 4 31B30.7B | 32.6 GB | Runs | 21.8 | 202K |
| Qwen3.6 35B A3B36B (3B active) | 36.9 GB | Runs | 121 | 256K |
RunsFits fully in VRAM; speed is estimated from VRAM bandwidth.
Needs RAM offloadMoE only: expert weights sit in system RAM, so speed is estimated from 70 GB/s RAM bandwidth.
Too bigWon’t fit on one card at this quantization, even with RAM offload.
Dual, 4x and 8x GeForce RTX 4090D 48GB for LLMs: which models fit?
Cards needed for 50B+ models released since 2026, with vLLM tensor parallelism at 32K context
| Model | Size | Cards | Total price | Total VRAM | Speed t/s | Total power |
|---|---|---|---|---|---|---|
| Qwen3.8 Flash Next176B (6B active) | 78.9 GB | 2 cards | $7,052 | 96 GB | 122.5 | 850 W |
 DeepSeek V4 Flash284B (13B active) | 96.8 GB | 4 cards | $14,104 | 192 GB | 99 | 1,700 W |
 Hy3295B (21B active) | 107.2 GB | 4 cards | $14,104 | 192 GB | 46.8 | 1,700 W |
| GLM 5.3 Flash320B (18B active) | 108.7 GB | 4 cards | $14,104 | 192 GB | 80 | 1,700 W |
| GLM 5.3744B (40B active) | 253.9 GB | 8 cards | $28,208 | 384 GB | 41 | 3,400 W |
| Hy4 preview770B (49B active) | 286.1 GBest. | 8 cards | $28,208 | 384 GB | 30.9 | 3,400 W |
 Kimi K2.7 Code1,000B (32B active) | 339.5 GB | 8 cards | $28,208 | 384 GB | 49.5 | 3,400 W |
| Model | Size | Cards | Total price | Total VRAM | Speed t/s | Total power |
|---|---|---|---|---|---|---|
| Qwen3.8 Flash Next176B (6B active) | 111.3 GB | 4 cards | $14,104 | 192 GB | 103.6 | 1,700 W |
| DeepSeek V4 Flash284B (13B active) | 155.1 GB | 4 cards | $14,104 | 192 GB | 73 | 1,700 W |
| Hy3295B (21B active) | 182.2 GB | 8 cards | $28,208 | 384 GB | 35 | 3,400 W |
| GLM 5.3 Flash320B (18B active) | 199.7 GB | 8 cards | $28,208 | 384 GB | 51.5 | 3,400 W |
| Model | Size | Cards | Total price | Total VRAM | Speed t/s | Total power |
|---|---|---|---|---|---|---|
| DeepSeek V4 Flash284B (13B active) | 161.9 GB | 4 cards | $14,104 | 192 GB | 70.8 | 1,700 W |
| Qwen3.8 Flash Next176B (6B active) | 188.2 GB | 8 cards | $28,208 | 384 GB | 75.8 | 3,400 W |
| Hy3295B (21B active) | 317.7 GB | 8 cards | $28,208 | 384 GB | 24 | 3,400 W |
| GLM 5.3 Flash320B (18B active) | 341 GB | 8 cards | $28,208 | 384 GB | 33.2 | 3,400 W |
Card counts are 1 / 2 / 4 / 8, each using 90% of its VRAM. Speed is based on one card’s bandwidth; price covers GPUs only. No NVLink; cards talk over PCIe. Whether it runs also depends on vLLM support.
Other markets
eBay
Amazon.jpBuy links may carry affiliate tags; they do not change the price you pay.
FAQ
How large a model can the GeForce RTX 4090D 48GB run?
A single GeForce RTX 4090D 48GB has 48GB of GDDR6X VRAM. At 4-bit quantization and 8K context it fits a dense model of up to about 79B parameters, or about 43B at 8-bit.
Can the GeForce RTX 4090D 48GB run a 70B model?
Yes. At 4-bit with an 8K context a 70B dense model needs about 43GB, which fits in the GeForce RTX 4090D 48GB's 48GB.
What are the best LLMs to run on the GeForce RTX 4090D 48GB?
Among models released in 2026, the largest dense model that fits entirely in VRAM at 4-bit is Gemma 4 31B, at an estimated 37.7 tokens/s. Among MoE models, Qwen3.6 35B A3B is the fastest at about 151.5 tokens/s. The full list is in the table above.
How many tokens per second does the GeForce RTX 4090D 48GB get on LLMs?
Estimated from memory bandwidth at 4-bit and 8K context: 8B dense: about 125.5 tokens/s, 14B dense: about 76.3 tokens/s, 32B dense: about 35.6 tokens/s, and 70B dense: about 16.9 tokens/s. MoE models read only the active parameters for each token, so they run much faster. Real-world results are usually 70%–100% of the estimate.
What LLMs can dual GeForce RTX 4090D 48GB cards run?
With vLLM tensor parallelism at 4-bit and 32K context, two GeForce RTX 4090D 48GB cards cannot fit any model of 50B or larger released in 2026. At 4-bit with 32K context the minimum is 4 cards, which fits Qwen3.8 Flash Next and DeepSeek V4 Flash.
How much does a GeForce RTX 4090D 48GB cost?
As of Sep 29, 2026: Xianyu median $3,526 (9 listings). Figures are medians of live listings, refreshed every 6 hours.
Is the GeForce RTX 4090D 48GB price going up or down?
As of Sep 29, 2026, the Xianyu median is down 2.5% over 7 days, down 1.3% over 30 days.
GeForce RTX 4090D 48GB vs RTX PRO 4500: which is better for LLMs?
The RTX PRO 4500 has 32GB of VRAM and 896 GB/s of bandwidth; the GeForce RTX 4090D 48GB has 48GB and 1,008 GB/s. At the Xianyu median, the RTX PRO 4500 costs about $3,622, 2.7% more than the GeForce RTX 4090D 48GB ($3,526). For 32B dense models at 4-bit, the GeForce RTX 4090D 48GB is estimated at about 35.6 tokens/s and the RTX PRO 4500 at about 31.8 tokens/s.