Price index · 2026-11
Vast.ai GPU prices — November 2026
Snapshot captured 09 Nov 2026, 18:14 UTC · live prices · JSON
Dated snapshot of the 45 GPU SKUs that had live whole-card offers on Vast.ai when it was taken in November 2026. The cheapest card was GTX 1660 Ti with 6 GB at $0.04/hour ($29/month at 24/7), median $0.08/hour.
Hourly and monthly rates in November 2026
| GPU | VRAM | Lowest $/h | Median $/h | Median $/month | Offers |
|---|---|---|---|---|---|
| GTX 1660 Ti | 6 GB | $0.04 | $0.08 | $58 | 5 |
| GTX 1660 S | 6 GB | $0.05 | $0.13 | $95 | 2 |
| RTX 3060 | 12 GB | $0.06 | $0.08 | $58 | 12 |
| GTX 1080 | 8 GB | $0.06 | $0.06 | $44 | 2 |
| GTX 1080 Ti | 11 GB | $0.07 | $0.07 | $51 | 3 |
| RTX 5060 | 8 GB | $0.08 | $0.09 | $66 | 2 |
| RTX 2080 Ti | 11 GB | $0.09 | $0.10 | $73 | 4 |
| RTX 3070 | 8 GB | $0.09 | $0.29 | $212 | 5 |
| L4 | 24 GB | $0.10 | $0.10 | $73 | 6 |
| RTX 5060 Ti | 16 GB | $0.10 | $0.17 | $124 | 9 |
| RTX A4000 | 16 GB | $0.10 | $0.10 | $73 | 3 |
| RTX 3060 Ti | 8 GB | $0.10 | $0.11 | $80 | 5 |
| RTX 3080 | 10 GB | $0.11 | $0.14 | $102 | 23 |
| RTX 5060 Ti | 8 GB | $0.11 | $0.12 | $88 | 4 |
| RTX 3090 | 24 GB | $0.12 | $0.20 | $146 | 14 |
| RTX 3080 Ti | 12 GB | $0.12 | $0.15 | $110 | 12 |
| RTX 4060 Ti | 8 GB | $0.12 | $0.15 | $110 | 4 |
| RTX 5070 | 12 GB | $0.13 | $0.17 | $124 | 8 |
| RTX 4070 | 12 GB | $0.13 | $0.14 | $102 | 3 |
| RTX 4090 | 24 GB | $0.14 | $0.42 | $307 | 33 |
| RTX 4060 Ti | 16 GB | $0.15 | $0.17 | $124 | 4 |
| RTX 5070 Ti | 16 GB | $0.16 | $0.24 | $175 | 19 |
| RTX A2000 | 6 GB | $0.16 | $0.22 | $161 | 2 |
| RTX 4000Ada | 20 GB | $0.18 | $0.20 | $146 | 3 |
| RTX 4070S Ti | 16 GB | $0.19 | $0.22 | $161 | 3 |
| RTX 3070 Ti | 8 GB | $0.19 | $0.19 | $139 | 1 |
| RTX 3090 Ti | 24 GB | $0.22 | $0.22 | $161 | 3 |
| RTX 4080S | 16 GB | $0.22 | $0.24 | $175 | 6 |
| RTX 5080 | 16 GB | $0.23 | $0.31 | $226 | 11 |
| RTX 4080 | 16 GB | $0.23 | $0.27 | $197 | 7 |
| RTX 4070 Ti | 12 GB | $0.23 | $0.23 | $168 | 1 |
| CMP 170HX | 64 GB | $0.24 | $0.24 | $175 | 1 |
| RTX PRO 4000 | 24 GB | $0.28 | $0.29 | $212 | 3 |
| RTX PRO 4500 | 32 GB | $0.35 | $0.40 | $292 | 3 |
| RTX A6000 | 48 GB | $0.40 | $0.51 | $372 | 4 |
| RTX 5090 | 32 GB | $0.43 | $0.56 | $409 | 28 |
| A100 SXM4 | 40 GB | $0.45 | $0.80 | $584 | 5 |
| RTX PRO 5000 | 48 GB | $0.51 | $0.64 | $467 | 2 |
| A100 PCIE | 40 GB | $0.54 | $0.64 | $467 | 2 |
| RTX 6000Ada | 48 GB | $0.62 | $0.67 | $489 | 3 |
| RTX 4090 | 48 GB | $0.68 | $1.34 | $978 | 3 |
| RTX PRO 6000 WS | 96 GB | $0.80 | $1.41 | $1,029 | 11 |
| RTX PRO 6000 Max-Q | 96 GB | $1.14 | $1.35 | $986 | 6 |
| RTX PRO 6000 S | 96 GB | $1.60 | $1.60 | $1,168 | 2 |
| H200 | 141 GB | $2.63 | $2.63 | $1,920 | 1 |
What these numbers mean
- Prices are what hosts were asking at capture time, not list prices.
- Only whole single cards count: shared slices and multi-GPU nodes never appear.
- A card enters the catalog once five or more whole single cards are offered at the same time; a snapshot keeps every card that still had at least one live offer when it was taken, so a lone leftover slice cannot set a floor.
- Two numbers per card: the cheapest single offer and the median across all offers — the median is what a typical rental costs.
- Monthly figures assume 730 hours (365 ÷ 12 × 24); Vast.ai bills per second.
The same data as JSON: /gpu-prices.json?period=2026-11.