joulemathGPU rental prices · Updated November 11, 2026
GPU rental prices per hour: H100, H200, B200, A100, RTX 4090
Short answer: on November 11, 2026 an H100 SXM rented for a median $2.67 an hour on Vast.ai (3 offers) and $2.69 on RunPod Community, an H200 for $3.55 to $5.29, a B200 for $5.98 to $9.69 and an RTX 4090 for $0.34 to $0.45. The Ornn index of real trades put the H100 at $3.10 and the H200 at $7.24 (settled November 10). Every price below is per GPU-hour, dated and linked.
How much does it cost to rent a GPU per hour?
| GPU | Vast.ai median, 1 GPU | RunPod Community | RunPod Secure | Lambda | Ornn index |
|---|---|---|---|---|---|
| B300 | Not checked | $6.94 | $8.99 | Not checked | Not checked |
| B200 | $9.69 (4 offers) | $5.98 | $7.99 | 1x $7.99, 8x $7.69 | $9.90 |
| H200 | $3.55 (7 offers) | $3.59 | $5.29 | Not checked | $7.24 |
| GH200 | Not checked | Not listed | Not listed | $2.59 | Not checked |
| H100 SXM | $2.67 (3 offers) | $2.69 | $3.99 | 1x $4.89, 8x $4.59 | $3.10 |
| H100 NVL | $2.67 (4 offers) | $2.59 | $3.19 | Not checked | Not checked |
| H100 PCIe | $2.47 (2 offers) | $1.99 | $2.89 | $3.79 | Not checked |
| A100 SXM | $0.72 (15 offers) | $1.39 | $1.79 | 8x 80 GB $3.19 | $1.05 |
| A100 PCIe | $0.67 (12 offers) | $1.19 | $1.79 | $1.99 | Not checked |
| RTX PRO 6000 | Not checked | $1.69 | $2.49 | Not checked | $1.56 |
| L40S | Not checked | $0.79 | $1.09 | Not checked | Not checked |
| RTX 5090 | $0.62 (64 offers) | $0.69 | $1.19 | Not listed | $0.68 |
| RTX 4090 | $0.45 (64 offers) | $0.34 | $0.89 | Not listed | Not checked |
| RTX 3090 | $0.21 (64 offers) | $0.22 | $0.50 | Not listed | Not checked |
What renters pay, USD per GPU-hour, read November 11, 2026. Vast.ai: our query of its public offers API at 11:03 CET, rentable, on-demand, exactly one GPU; offer counts stop at 64. RunPod: pricing page dated November 9, 2026. Lambda: on-demand, plus tax; published list says prices take effect November 16, 2026; it also lists an A100 40 GB at $1.99. Ornn: OCPI settlement prices for November 10, 2026 (free-tier index, including a new RTX 5090 index at $0.68). "Not checked" means we did not read that price, not that it does not exist.
Which GPUs have full pages?
- NVIDIA H100 rental price per hour: SXM vs PCIe specs, Ornn, RunPod, Lambda and Vast.ai prices, and profit per kW.
- NVIDIA H200 rental price per hour: RunPod, Vast.ai, Nebius and Ornn prices, why the trade index runs higher, H200 vs H100, and profit per kW.
- NVIDIA A100 rental price per hour: SXM vs PCIe specs, Ornn, RunPod, Lambda and Vast.ai prices, and profit per kW.
- NVIDIA RTX 4090: Vast.ai, RunPod and SaladCloud prices, used card prices, power cost and owner payback.
Own a gaming card? What renting out an RTX 4090, 5090 or 3090 earns after power. Or compare against mining in the per-kW calculator, and see what the same kilowatts earn on the ASIC pages.
Latest GPU price notes
- H100 price per hour across clouds (October 26, 2026)
- GPU rental prices: H100, H200, A100 and RTX 4090 (October 2026)
- Hashprice vs GPU rent: what 1 kW earns
- Bitcoin miners pivot to AI hosting: the deals and why
FAQ
How much does it cost to rent an H100 per hour?
On November 11, 2026 one H100 SXM rented for a median $2.67 an hour on Vast.ai (3 offers), $2.69 on RunPod Community, $3.99 on RunPod Secure and $4.89 on Lambda (1x GPU, published list). The Ornn index of executed trades settled at $3.10 on November 10.
What is the cheapest GPU to rent?
Of the cards in the table, the RTX 3090: a median $0.21 an hour on Vast.ai and $0.22 on RunPod Community on November 11, 2026. For data-center GPUs, an A100 SXM had a Vast.ai median of $0.72 and RunPod Community listed $1.39.
How much does a B200 cost to rent?
RunPod listed the B200 at $5.98 an hour on Community Cloud and $7.99 on Secure Cloud (page dated November 9, 2026). Lambda listed $7.99 for 1x and $7.69 per GPU for 8x (published list; page says those take effect November 16, 2026). The Ornn B200 index settled at $9.90 on November 10. Vast.ai had 4 single-GPU on-demand offers, $7.50 to $12.51, median $9.69, on November 11.
Why do GPU rental prices differ so much between providers?
Vast.ai is a marketplace where hosts set their own prices, so rates move with supply and offer quality. RunPod and Lambda post fixed on-demand prices, and RunPod Secure Cloud runs in data centers, which costs more than Community Cloud. Ornn is an index of executed trades, not a list price.
Can I rent out my own GPU?
Yes, if it is a gaming card on Salad, or a dedicated Linux box on Vast.ai. RunPod Community Cloud is closed to new hosts. What a 4090, 5090 or 3090 keeps after power is in our guide to renting out your GPU.
Already hosting GPUs? Benchmark a Vast.ai machine in Fleet Check.
How do Vast.ai hosts track earnings and payments?
Connect your Vast.ai account to Flash to track earnings, fees and payouts by machine. The integration is available to hosts today, and Flash can help with payments as well as sync to QuickBooks in one click.
Run GPUs for AI?
Connect Vast.ai to Flash to track earnings, fees and payouts by machine. Get help with payments and sync to QuickBooks in one click.