24h drops
A100 80GB PCIe$17,842-20.0%
Arc Pro B70$1,443-10.2%- Tesla V100 32GB$554-9.1%
- GeForce RTX 5060 Ti 16GB$714-5.2%
- Mac Studio M3 Ultra 512GB$16,142-5.0%
24h rises
- RTX PRO 5000 48GB$6,992+17.5%
- GeForce RTX 4090$3,719+8.7%
- Radeon PRO W7900$3,050+4.1%
- GeForce RTX 4080$1,190+3.9%
- GeForce RTX 5070 Ti$1,264+3.9%
Most listings
- Mac Studio M5 Max 128GB$6,10035 listings
- GeForce RTX 5070$92629 listings
- Radeon RX 7900 XTX$88128 listings
- Radeon RX 6950 XT$45826 listings
- Radeon RX 9070 XT$78821 listings
Latest release
DeepSeek V4.1 Flash763BQ4444.7 GB
Hy4 preview770BQ4467.3 GB
GLM 5.3744BQ4467.3 GB- GLM 5.3 Flash320BQ4199.7 GB
Qwen3.8 Flash Next176BQ4111.3 GB
Local LLM GPU Ranking
Xianyu
| Model | 30d trend | Ecosystem | |||||||
|---|---|---|---|---|---|---|---|---|---|
| H100 PCIe | No price in this market | — | 80 | 2,000 | 756 | 1,513 | 350 | CUDA | |
| RTX PRO 6000 | $18,2993 listings | -4.6% | 96 | 1,792 | 500 | 1,000 | 600 | CUDA | |
| GeForce RTX 5090 | $6,4966 listings | -2.3% | 32 | 1,792 | 419 | 838 | 575 | CUDA | |
| RTX PRO 6000D | $9,7457 listings | +1.6% | 84 | 1,398 | 385.2 | 770.5 | 600 | CUDA | |
| L40S | No price in this market | — | 48 | 864 | 362.1 | 733 | 350 | CUDA | |
| RTX 6000 Ada | $9,74511 listings | -0.4% | 48 | 960 | 364.3 | 728.5 | 300 | CUDA | |
| GeForce RTX 4090 | $3,7195 listings | +8.7% | 24 | 1,008 | 330.3 | 660.6 | 450 | CUDA | |
| GeForce RTX 4090 48GB | No price in this market | — | 48 | 1,008 | 330.3 | 660.6 | 450 | CUDA | |
| A100 40GB PCIe | $4,6863 listings | -1.6% | 40 | 1,555 | 312 | 624 | 250 | CUDA | |
| A100 80GB PCIe | $17,8424 listings | -20.0% | 80 | 1,935 | 312 | 624 | 300 | CUDA | |
| GeForce RTX 5090 D | $4,76111 listings | 0.0% | 32 | 1,792 | 296.9 | 593.8 | 575 | CUDA | |
| GeForce RTX 5090 D V2 | $3,7199 listings | +0.2% | 24 | 1,344 | 296.9 | 593.8 | 575 | CUDA | |
| GeForce RTX 4090D 48GB | $3,5269 listings | -1.7% | 48 | 1,008 | 294.3 | 588.5 | 425 | CUDA | |
| RTX PRO 5000 48GB | $6,9927 listings | +17.5% | 48 | 1,344 | 258 | 516 | 300 | CUDA | |
| RTX PRO 5000 72GB | $9,52214 listings | +1.6% | 72 | 1,344 | 258 | 516 | 300 | CUDA | |
| GeForce RTX 5080 | $1,75515 listings | +0.1% | 16 | 960 | 225.1 | 450.2 | 360 | CUDA | |
| GeForce RTX 4080 SUPER 32GB | No price in this market | — | 32 | 736 | 208.9 | 417.8 | 320 | CUDA | |
| RTX PRO 4500 | $3,6228 listings | 0.0% | 32 | 896 | 202.1 | 404.3 | 200 | CUDA | |
| CMP 170HX 40GB | $1,7853 listings | 0.0% | 40 | 1,560 | 202 | 404 | 250 | CUDA | |
| CMP 170HX 64GB | $2,1575 listings | 0.0% | 64 | 1,493 | 202 | 404 | 250 | CUDA | |
| GeForce RTX 4080 | $1,1906 listings | +3.9% | 16 | 717 | 195 | 389.9 | 320 | CUDA | |
| Radeon RX 9070 XT | $78821 listings | +1.9% | 16 | 640 | 195 | 389 | 304 | ROCm | |
| Radeon AI PRO R9700 | $1,75619 listings | +0.2% | 32 | 640 | 191 | 383 | 300 | ROCm | |
| Arc Pro B70 | $1,44310 listings | -10.2% | 32 | 608 | 183.5 | 367 | 230 | oneAPI | |
| GeForce RTX 5070 Ti | $1,26416 listings | +3.9% | 16 | 896 | 175.8 | 351.5 | 300 | CUDA | |
| RTX PRO 4000 | $2,23213 listings | 0.0% | 24 | 672 | 161.3 | 322.5 | 145 | CUDA | |
| RTX A6000 | $4,46313 listings | -2.9% | 48 | 768 | 154.9 | 309.7 | 300 | CUDA | |
| GeForce RTX 3090 | $1,20510 listings | -1.2% | 24 | 936 | 142.3 | 284.7 | 350 | CUDA | |
| DGX Spark 128GB | $5,09614 listings | -2.1% | 128 | 273 | 125 | 250 | 240 | CUDA | |
| GeForce RTX 5070 | $92629 listings | +2.1% | 12 | 672 | 123.5 | 247 | 250 | CUDA | |
| GeForce RTX 2080 Ti 22GB | $3943 listings | 0.0% | 22 | 616 | 107.6 | 215.2 | 250 | CUDA | |
| Arc Pro B60 | $8264 listings | +2.8% | 24 | 456 | 98.5 | 197 | 200 | oneAPI | |
| Arc Pro B65 | $1,2423 listings | 0.0% | 32 | 608 | 98.5 | 197 | 200 | oneAPI | |
| GeForce RTX 5060 Ti 16GB | $71411 listings | -5.2% | 16 | 448 | 94.9 | 189.8 | 180 | CUDA | |
| Instinct MI210 | $2,7523 listings | 0.0% | 64 | 1,638 | 181 | 181 | 300 | ROCm | |
| GeForce RTX 4060 Ti 16GB | $5373 listings | +0.3% | 16 | 288 | 88.3 | 176.5 | 165 | CUDA | |
| Tesla T10 | $1999 listings | +2.7% | 16 | 403 | 80 | 160 | 150 | CUDA | |
| Radeon PRO W7900 | $3,0503 listings | +4.1% | 48 | 864 | 123 | 123 | 295 | ROCm | |
| Radeon RX 7900 XTX | $88128 listings | -0.5% | 24 | 960 | 123 | 123 | 355 | ROCm | |
| Radeon RX 7900 XT | $62518 listings | 0.0% | 20 | 800 | 103 | 103 | 315 | ROCm | |
| GeForce RTX 3060 12GB | $2958 listings | -0.7% | 12 | 360 | 51 | 101.9 | 170 | CUDA | |
| Radeon RX 6950 XT | $45826 listings | +2.7% | 16 | 576 | 47.3 | 94.6 | 335 | ROCm | |
| Instinct MI100 | No price in this market | — | 32 | 1,229 | 92.3 | 92.3 | 300 | ROCm | |
| Radeon PRO W7800 | $1,8583 listings | +0.7% | 32 | 576 | 90.4 | 90.4 | 260 | ROCm | |
| Radeon RX 7800 XT | $46116 listings | +0.3% | 16 | 624 | 74.7 | 74.7 | 263 | ROCm | |
| Ryzen AI Max+ 395 128GB | $2,65721 listings | -0.2% | 128 | 256 | 59.4 | 59.4 | 120 | ROCm | |
| Tesla V100 16GB | $1795 listings | -4.8% | 16 | 900 | 112 | 56.1 | 250 | CUDA | |
| Tesla V100 32GB | $5548 listings | -9.1% | 32 | 900 | 112 | 56.1 | 250 | CUDA | |
| Instinct MI50 32GB | $5066 listings | 0.0% | 32 | 1,024 | 26.5 | 53 | 300 | ROCm | |
| Tesla P40 | $17913 listings | -1.6% | 24 | 347 | — | 47 | 250 | CUDA | |
| Mac Studio M3 Ultra 512GB | $16,14212 listings | -5.0% | 512 | 819 | — | — | 480 | Metal | |
| Mac Studio M4 Max 128GB | $5,04311 listings | -3.1% | 128 | 546 | — | — | 480 | Metal | |
| Mac Studio M5 Max 128GB | $6,10035 listings | -0.1% | 128 | 614 | — | — | — | Metal | |
| MacBook Pro M4 Max 128GB | $5,2079 listings | 0.0% | 128 | 546 | — | — | 140 | Metal | |
| MacBook Pro M5 Max 128GB | $7,14113 listings | -1.4% | 128 | 614 | — | — | 140 | Metal | |
| RTX PRO 5500 | No price in this market | — | 84 | 1,398 | — | — | 600 | CUDA |
No GPUs match “”.
Log in to add GPUs to your watchlist and see it on any device. Log in
No GPUs in your watchlist yet. Click the ☆ before a model name to add one.
DDR4 RAM prices
Xianyu
| Spec | Price per stick | Price per GB | Max channels | Max bandwidth | Max memory |
|---|---|---|---|---|---|
| 16 GB | $7412 listings | $4.65 | 2 | 51.2 GB/s | 64 GB4 sticks × 16 GB |
| 32 GB | $1504 listings | $4.7 | 2 | 51.2 GB/s | 128 GB4 sticks × 32 GB |
| 16 GB | $1199 listings | $7.4 | 8 | 170.6 GB/s | 128 GB8 sticks × 16 GB |
| 32 GB | $2385 listings | $7.4 | 8 | 170.6 GB/s | 256 GB8 sticks × 32 GB |
DDR5 RAM prices
Xianyu
| Spec | Price per stick | Price per GB | Max channels | Max bandwidth | Max memory |
|---|---|---|---|---|---|
| 16 GB | $19825 listings | $12.4 | 2 | 96 GB/s | 64 GB4 sticks × 16 GB |
| 32 GB | $4414 listings | $13.8 | 2 | 96 GB/s | 128 GB4 sticks × 32 GB |
| 32 GB | $1,19012 listings | $37.2 | 12 | 537.6 GB/s | 384 GB12 sticks × 32 GB |
| 64 GB | $2,3066 listings | $36 | 12 | 537.6 GB/s | 768 GB12 sticks × 64 GB |
Choose hardware for local LLM deployment
Find the right GPU for local LLMs. Compare VRAM, GPU prices and model memory requirements to choose hardware that fits your models and budget.
What is VRAM, and how much do you need?
VRAM is the fast memory on a graphics card. Local LLM inference needs space for model weights, the context (KV) cache and runtime buffers. Quantization can reduce weight memory, while longer prompts and more simultaneous users need additional memory. A model fitting in VRAM does not guarantee a particular speed.
Plan a local deployment
Choose a model and quantization, set the context length, and check both GPU memory and system RAM. CPU or expert offloading can make a large model fit with less VRAM, but it depends on RAM bandwidth and can be much slower. Check CUDA, ROCm or Metal support before buying hardware.
Read the rankings and price data
Compare VRAM capacity first, then bandwidth, power and software support. Our prices are median asking prices, not completed sales; sample counts, observation dates and stale-price labels help you judge their usefulness. INT8 throughput alone is not an LLM benchmark.
