GPUhub In stock
Singapore 4 configs 1x-8x
China-market export variant of the RTX PRO 6000 for local AI and visualization.
The RTX PRO 6000D is available at 1 cloud provider for $0.82/hr per GPU.
Singapore 4 configs 1x-8x
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.
84 GB GDDR7 Blackwell GPU sold in the Chinese market, where the full RTX PRO 6000 is export-restricted. Suited to AI inference workloads.
Export-compliance variant with 17% fewer CUDA cores, 84 GB (vs 96 GB), ~1,398 GB/s bandwidth, and no NVLink versus the standard RTX PRO 6000.
The median on-demand price has held steady around $0.82/hr per GPU across providers.
With 84GB of VRAM, the RTX PRO 6000D is well suited to 30B-class models in FP16, and 70B-class models in 4-bit or 8-bit quantized form.
The RTX PRO 6000D has 84GB of VRAM. Multi-GPU setups increase total memory, but that memory is not automatically pooled across GPUs.
The RTX PRO 6000D has 1,568 GB/s of memory bandwidth. Higher bandwidth helps with faster data transfer between GPU memory and compute cores.
The RTX PRO 6000D supports 7 precision formats. Training: BF16, FP16, TF32, FP32. Inference: FP4, FP8, INT8.
No. The RTX PRO 6000D is a PCIe-only GPU with no NVLink, so it is better suited to single-GPU inference and smaller-scale workloads than large distributed training jobs.
RTX PRO 6000D pricing is currently $0.82/hr per GPU.
At 720 hours per month, one RTX PRO 6000D currently costs about $588.24/mo.
The RTX PRO 6000D is available from 1 cloud provider: GPUhub.
Yes. We currently track 4 RTX PRO 6000D listings across 1 cloud providers:
| Billing type | Listings | Avg $/GPU/hr |
|---|---|---|
| On-demand | 4 | $0.82/hr |
| GPU architecture | NVIDIA Blackwell |
| NVIDIA® CUDA® Cores | 19,968 |
| Tensor Cores | 5th Generation |
| Ray Tracing Cores | 4th Generation |
| Boost Clock | 2.43 GHz |
| GPU memory | 84 GB GDDR7 |
| Memory interface | 448-bit |
| Memory bandwidth | ~1568 GB/s (estimated) |
| System interface | PCIe 5.0 x16 |
| NVLink Support | No |
| Market | China only (US export-control compliant) |
Export-compliant variant of the RTX PRO 6000. Reduced CUDA cores (19,968 vs 24,064), memory (84 GB vs 96 GB), and clocks keep it under US export thresholds. Targets AI inference rather than large-scale training.
Note: Nvidia does not publish an official datasheet for this China-only variant. Specs have been drawn from a board teardown by UNIKO's Hardware. The 84 GB capacity and 448-bit bus are estimated by chip count (28 × 3 GB GDDR7), and memory bandwidth is an estimate assuming 28 Gbps memory.
Standard global Blackwell workstation GPU with more memory and cores. From $0.34/hr per GPU across 34 providers.
Consumer Blackwell GPU with less memory. From $0.08/hr per GPU across 14 providers.
Hopper data center GPU sold as a China-market export part.
Ada Lovelace professional card for rendering and AI development. From $0.14/hr per GPU across 12 providers.
Last updated