Sold prices / best / local AI
Best used GPU for local AI
The cheapest VRAM in a used desktop card with 12GB or more right now is the RTX 2060 12GB: 12GB for a median $223 on US eBay, or $19 per GB. The cheapest 24GB card, enough for 32B models at 4-bit, is the Quadro RTX 6000 at $900. Prices as of 2026-10-31.
RTX 2070: up 4% over the past week.
Cheapest card at each VRAM size
VRAM decides how big a model fits. Speed matters less than fitting at all.
| VRAM | Card | Sold median | Runs at 4-bit |
|---|---|---|---|
| 8GB+ | RTX 2070Or Tesla P4 8GB at $115, a fanless server card | $187 | 8B models |
| 12GB+ | RTX 2060 12GBOr Tesla P100 16GB at $140, a fanless server card | $223 | 14B models |
| 16GB+ | RTX 4060 Ti 16GBOr Tesla P100 16GB at $140, a fanless server card | $631 | 14B models |
| 24GB+ | Quadro RTX 6000Or Tesla P40 24GB at $293, a fanless server card | $900 | 32B models |
| 32GB+ | RTX 5000 AdaOr Tesla V100 32GB at $855, a fanless server card | $2,825 | 32B models |
| 48GB+ | Quadro RTX 8000 | $2,070 | 70B models |
All NVIDIA cards by price per GB of VRAM
| Card | VRAM | Sold median | Per GB |
|---|---|---|---|
| Tesla P100 16GB25 sales · server card, no fan | 16GB | $140 | $9 |
| Tesla P40 24GB26 sales · server card, no fan | 24GB | $293 | $12 |
| Tesla P4 8GB13 sales · server card, no fan | 8GB | $115 | $14 |
| RTX 2060 12GB14 sales | 12GB | $223 | $19 |
| Tesla V100 16GB20 sales · server card, no fan | 16GB | $329 | $21 |
| RTX 207066 sales | 8GB | $187 | $23 |
| RTX 2060 Super48 sales | 8GB | $190 | $24 |
| RTX 3050 8GB63 sales | 8GB | $200 | $25 |
| RTX 2070 Super110 sales | 8GB | $200 | $25 |
| RTX 208067 sales | 8GB | $208 | $26 |
| Tesla V100 32GB29 sales · server card, no fan | 32GB | $855 | $27 |
| RTX 2080 Ti100 sales | 11GB | $299 | $27 |
| RTX 3060 12GB284 sales | 12GB | $330 | $28 |
| Tesla T4 16GB25 sales · server card, no fan | 16GB | $445 | $28 |
| RTX 2080 Super57 sales | 8GB | $229 | $29 |
| RTX 3060 8GB27 sales | 8GB | $250 | $31 |
| RTX 3060 Ti205 sales | 8GB | $279 | $35 |
| RTX 3070244 sales | 8GB | $296 | $37 |
| Quadro RTX 600011 sales | 24GB | $900 | $38 |
| Titan RTX32 sales | 24GB | $904 | $38 |
| RTX 4060 Ti 16GB30 sales | 16GB | $631 | $39 |
| RTX 3080 12GB17 sales | 12GB | $499 | $42 |
| RTX A2000 12GB23 sales | 12GB | $500 | $42 |
| RTX 3080 10GB192 sales | 10GB | $420 | $42 |
| RTX 3070 Ti88 sales | 8GB | $339 | $42 |
NVIDIA only, because CUDA is what most local AI tools support best. Older Tesla and Pascal cards are cheap per GB but slow and need extra cooling. Model sizes assume 4-bit weights with room for context.
Complete local AI builds with every part priced
Prices are medians of completed US eBay sales over the last 14 days (30 when sales are thin), shipping included, with broken, parts-only, and bundle listings removed. Cards with too few sales or a wide spread are left out. How we price