Two Nvidia workstation GPUs on the Pascal architecture: 8 GB GDDR5 against 24 GB GDDR5X and 78% more memory bandwidth on the P6000.
At a glance
Updated 18 minutes ago- Memory
- 8 GB
- GDDR5
- Memory bandwidth
- 243 GB/s
- From
- $0.05 /GPU/hr
- 2 providers
- Memory
- 24 GB
- GDDR5X
- Memory bandwidth
- 432 GB/s
- From
- $1.10 /GPU/hr
- 1 provider
Price comparison
Cheapest listed price per GPU per hour by billing type, and the median across providers.
| Billing | P4000 | P6000 | Difference |
|---|---|---|---|
| On-demand | $0.05 | $1.10 | 20x more expensive on P6000 |
| Spot | $0.04 | No spot listings | |
| Reserved | $0.06 | No reserved listings | |
| Median | $0.28 | $1.10 | 3.9x more expensive on P6000 |
Median: Median across providers, one vote each, so a large catalog does not outweigh a small one.
Price history
P4000 vs P6000 price history
Weekly median on-demand price per GPU per hour
Specs comparison
| Spec | P4000 | P6000 | Difference |
|---|---|---|---|
| Memory | |||
| Capacity | 8 GB GDDR5 | 24 GB GDDR5X | 3x more on P6000 |
| Bandwidth | 243 GB/s | 432 GB/s | 1.8x higher on P6000 |
| Bus width | 256-bit | 384-bit | 1.5x wider on P6000 |
| Compute · dense | |||
| FP32 | 5.3 TFLOPS | 12.6 TFLOPS | 2.4x higher on P6000 |
| Platform | |||
| Interconnect | PCIe only | PCIe only | |
| Board power | 105 W | 250 W | 2.4x higher on P6000 |
| Released | Q1 2017 | Q4 2016 | |
| Precision | |||
| Supported formats | FP16FP32 | FP16FP32 | |
Compute: Vendor peak figures, dense and per GPU. Not measured throughput.
Source: P4000 datasheet and P6000 datasheet.
Where to rent them
Providers listing these GPUs, those with both first, so you can switch without moving clouds.
| Provider | P4000 /GPU/hr | P6000 /GPU/hr |
|---|---|---|
Paperspace
|
On-Demand $0.51 | On-Demand $1.10 |
Vast.ai
|
On-Demand $0.05 | -- |
Similar GPUs
Other comparisons in the same class.
Our data for Nvidia P4000 was last updated on Sept. 7, 2026, and for Nvidia P6000 on Sept. 7, 2026.