Legacy Pascal data center GPU for budget scientific computing.
For smaller projects, the P100 is available around $0.20/hr per GPU (on-demand). This tier offers a 90% discount compared to the higher end of the market ($2.22/hr). Spot instances start lower, at $0.13/hr per GPU.
USA 4 configs 1x-2x
USA 36 configs 1x-4x
USA 2 configs 1x-2x
France 1 config 1x
Singapore 12 configs 1x-8x
USA 4 configs 1x
Heads up: We do our best to keep these specs & prices accurate. However, cloud costs may fluctuate based on region, usage, and other factors not listed here. These are estimates based on common setups and are for informational purposes only. Always verify current rates & exact specs with the provider before provisioning.
16GB HBM2 with FP16 and FP64 support. Very low cost per hour. Useful for legacy workloads, scientific computing, and basic GPU development.
Pascal architecture, 3+ generations behind. No Tensor Cores, no BF16, no INT8. Only viable for legacy FP32/FP64 scientific workloads or basic GPU development.
The median on-demand price across providers has risen about 11% since August 2025, from $1.28 to $1.42/hr per GPU. This reflects the market-wide median, which moves both when providers change prices and when lower or higher priced offerings enter the market.
With 16GB of VRAM, the P100 is best for 7B-class models in 4-bit or 8-bit quantized form, and smaller models in FP16.
The P100 has 16GB of VRAM. Multi-GPU setups increase total memory, but that memory is not automatically pooled across GPUs.
The P100 has 732 GB/s of memory bandwidth. Higher bandwidth helps with faster data transfer between GPU memory and compute cores.
The P100 supports 3 precision formats. Training: FP16, FP32. Scientific: FP64.
Yes. The P100 supports NVLink with 160 GB/s of bidirectional bandwidth. This helps accelerate multi-GPU communication.
P100 pricing currently ranges from $0.12/hr to $2.22/hr per GPU, depending on the provider, instance type, and billing model.
At 720 hours per month, one P100 can cost between $89.50 to $1,598.40 per month, depending on the provider. Reserved and spot pricing can lower that further.
The P100 is available from 6 cloud providers, including Scaleway, Alibaba Cloud, Database Mart. Pricing and availability vary by region and billing model.
Yes. We currently track 59 P100 listings across 6 cloud providers:
| Billing type | Listings | Avg $/GPU/hr |
|---|---|---|
| On-demand | 20 | $1.52/hr |
| Reserved | 28 | $1.05/hr |
| Spot | 11 | $0.17/hr |
| GPU Architecture | NVIDIA Pascal |
| NVIDIA CUDA® Cores | 3,584 |
| Double-Precision Performance | 4.7 TeraFLOPS |
| Single-Precision Performance | 9.3 TeraFLOPS |
| Half-Precision Performance | 18.7 TeraFLOPS |
| GPU Memory | 16GB CoWoS HBM2 at 732 GB/s or 12GB CoWoS HBM2 at 549 GB/s |
| System Interface | PCIe Gen3 |
| Max Power Consumption | 250 W |
| ECC | Yes |
| Thermal Solution | Passive |
| Form Factor | PCIe Full Height/Length |
| Compute APIs | CUDA, DirectCompute, OpenCL™, OpenACC |
Source: official Nvidia P100 datasheet.
Last updated