NVIDIA RTX Pro 4000 Blackwell
Product Overview
The NVIDIA RTX Pro 4000 Blackwell is a professional workstation GPU launched by NVIDIA in 2025 based on the Blackwell architecture, positioned as the entry model of the RTX Pro series. With 24GB GDDR7 ECC memory and a 145W TDP (plus an SFF single-slot 70W version), it retains 5th-generation Tensor Cores and FP4 inference capability, targeting lightweight AI development and professional graphics.
Core Specifications
| Parameter | Value |
|---|
| Architecture | NVIDIA Blackwell 2.0 (GB202) |
| Process | TSMC 4nm |
| Memory | 24 GB GDDR7 ECC |
| Memory Type | GDDR7 |
| Memory Bandwidth | Not disclosed (estimated ~600 GB/s) |
| FP32 | 40 TFLOPS |
| FP4 Tensor Core | 1,290 TOPS (sparse) |
| FP16/BF16 Tensor Core | 322 TFLOPS (sparse) |
| TDP | 145 W (SFF version: 70W) |
| PCIe | Gen 5 x16 |
| ECC | Supported |
| MIG | Supported |
| Release | 2025 |
Variant Differences
| Variant | TDP | Form Factor |
|---|
| Workstation edition | 145W | Dual-slot active cooling |
| SFF edition | 70W | Single-slot compact |
Key Features
- 24GB GDDR7 ECC memory: local inference of 14B-30B models
- 5th Gen Tensor Cores: FP4 inference support
- Professional drivers + ECC: the choice for workstation stability
- 145W low power: SFF version at 70W fits small chassis
Use Cases
- ✅ Lightweight AI development workstations
- ✅ Local LLM inference (14B-30B)
- ✅ Professional graphics
- ✅ Small workstations / edge nodes
External Links