NVIDIA RTX PRO 4500 Blackwell Workstation Edition 32GB Specs, Benchmarks & Pricing

The NVIDIA RTX PRO 4500 Blackwell Workstation Edition is a professional dual-slot workstation GPU launched on March 18, 2025, built on the NVIDIA Blackwell architecture (GB203 die, TSMC 4nm process) with 10,496 CUDA cores across 82 streaming multiprocessors, 328 fifth-generation Tensor Cores, and 82 fourth-generation RT Cores. It is positioned above the RTX PRO 4000 Blackwell in NVIDIA's professional desktop lineup and delivers 51 TFLOPS of single-precision FP32 compute and 1,617 AI TOPS (theoretical FP4 with sparsity).

The card features 32 GB of GDDR7 ECC memory on a 256-bit memory bus providing 896 GB/s bandwidth. The fifth-generation Tensor Cores introduce FP4 precision support alongside FP8, BF16, TF32, FP16, INT8, and INT4, enabling broad AI and machine learning workloads on professional applications. The fourth-generation RT Cores provide hardware-accelerated ray tracing with doubled throughput versus the prior generation.

With a TDP of 200W in a dual-slot, full-height form factor (4.4" H × 10.5" L), the RTX PRO 4500 Blackwell Workstation Edition requires one 16-pin PCIe CEM5 power connector. It connects via PCIe 5.0 x16 and provides four DisplayPort 2.1b outputs supporting up to 8K at 240 Hz or 16K at 60 Hz. Video engine support includes two ninth-generation NVENC encoders and two sixth-generation NVDEC decoders, a step up from the single-codec configuration of the RTX PRO 4000.

The RTX PRO 4500 Blackwell uses the GB203 die (the same die as the GeForce RTX 5080 which has all 10,752 shaders enabled and the RTX PRO 4000 with 8,960 shaders), with all 82 SMs active from the full 84-SM configuration of the die. The GPU is distributed through PNY and available from workstation OEMs including Dell and HP.

Strengths

  • Excellent FP16 compute performance (top 20% of GPUs)
  • Excellent memory capacity (top 20% of GPUs)

Specifications for NVIDIA RTX PRO 4500 Blackwell Workstation Edition

SpecificationPerformance Ranking
FP32 TFLOPs
68th @ 51 TFLOPs (Mid Tier)(Mid)
FP16 TFLOPs
80th @ 404.25 TFLOPs (Top Tier)(Top)
Tensor Core Count
71st @ 328 Cores (Mid Tier)(Mid)
Memory Capacity (GB)
80th @ 32 GB (Top Tier)(Top)
Memory Bandwidth (GB/s)
74th @ 896 GB/s (Mid Tier)(Mid)
Int8 TOPs
71st @ 404.25 TOPs (Mid Tier)(Mid)

Real-time NVIDIA RTX PRO 4500 Blackwell Workstation Edition GPU Prices

We're tracking 0 of the NVIDIA RTX PRO 4500 Blackwell Workstation Edition GPUs currently available for sale.
Buy Now

Compare Price/Performance to other GPUs

We track real-time prices of other GPUs too so that you can compare the price/performance of the NVIDIA RTX PRO 4500 Blackwell Workstation Edition GPU to other GPUs.
Compare GPU Price/Performance

Compare NVIDIA RTX PRO 4500 Blackwell Workstation Edition to Another GPU

Compare the NVIDIA RTX PRO 4500 Blackwell Workstation Edition directly to another GPU to see specs, benchmarks, and prices side-by-side.
Compare GPUs Side-by-Side

Price History

NVIDIA RTX PRO 4500 Blackwell Workstation Edition Price History

Insufficient historical data for price trends. More data will be available as we continue tracking prices.

Product Identifiers

Manufacturer Part Numbers (2)
NVIDIA Part Number
900-5G147-2250-000
NVIDIA Part Number
900-5G147-2550-000
Available from 3 Partners (3 products)
PNY
NVIDIA RTX PRO 4500 Blackwell Workstation Edition 32GB GDDR7 Graphics Card
VCNRTXPRO4500B-PB(sku)
Dell
Dell NVIDIA RTX PRO 4500 Blackwell Workstation Edition, 32 GB GDDR7, full height, PCIe 5.0x16, 4 DP
490-BLPL(part number)
HP
HP NVIDIA RTX PRO 4500 Blackwell 32 GB 4DP Graphics Card
B11F2AA(part number)

References

Notes

  1. FP32 TFLOPS of 51.0 sourced directly from the PNY official datasheet for the RTX PRO 4500 Blackwell Workstation Edition (pny.com), footnote 1 states it is based on GPU Boost clock. This is consistent with 10,496 CUDA cores at the TechPowerUp-reported 2407 MHz boost clock: 10,496 × 2407 × 2 / 1,000,000 ≈ 50.53 TFLOPS ≈ 51 TFLOPS. CG Channel reported 39.6 TFLOPS in their March 2025 announcement article; the PNY datasheet is more authoritative and the 51 TFLOPS figure is consistent with the known boost clock.
  2. AI TOPS of 1,617 is from the PNY official datasheet footnotes, labeled as theoretical FP4 TOPS using the sparsity feature (footnote 2 confirms sparsity is applied). This follows the same convention as the RTX PRO 4000 Blackwell where "AI TOPS" = FP4 sparse.
  3. int8TOPS of 404.25 represents non-sparse (dense) INT8 performance, derived from FP4 sparse TOPS: 1,617 (FP4 sparse) / 4 = 404.25 TOPS. Derivation: FP4 sparse = 2x INT8 sparse = 4x INT8 dense per Blackwell 5th-gen tensor core architecture. Dense value used for consistent cross-vendor comparison.
  4. fp16TFLOPS of 404.25 is FP16 Tensor Core throughput with 2:4 structured sparsity, derived from FP4 sparse TOPS using the Blackwell tensor core ratio (FP16 sparse = FP4 sparse / 4): 1,617 / 4 = 404.25 TFLOPS. Dense FP16 Tensor = 404.25 / 2 = 202.125 TFLOPS. The NVIDIA product page does not publish FP16 separately for this workstation card.
  5. Memory bandwidth of 896 GB/s per NVIDIA official product page and PNY datasheet. The memory is 32 GB GDDR7 ECC on a 256-bit bus. TechPowerUp reports 1750 MHz memory clock; 256 bits / 8 bytes × 1750 MHz × 2 (DDR) / 1000 = 896 GB/s.
  6. CUDA core count of 10,496 confirmed by TechPowerUp GPU database meta (GB203, 10496 Cores), PNY datasheet, and Newegg product listings. TechPowerUp also confirms 328 TMUs and 112 ROPs. GPU die is GB203 (same as GeForce RTX 5080 which has all 10,752 shaders enabled); the RTX PRO 4500 Blackwell has 10,496 shaders with 82 of 84 SMs active.
  7. Tensor core count of 328 confirmed by Newegg product listing for NVPN 900-5G147-2250-000 (specifications table states "Tensor Cores: 328") and matches the SM count: 82 SMs × 4 tensor cores per SM = 328 tensor cores. RT core count of 82 (one per SM, 4th generation).
  8. TDP of 200W per NVIDIA official product page and PNY datasheet. Requires one 16-pin PCIe CEM5 power connector per PNY datasheet and Newegg listing.
  9. GPU die is GB203 on TSMC 4nm process. TechPowerUp meta tag confirms GB203 die. Architecture uses compute capability 12.0 (sm_120), confirmed by the official NVIDIA CUDA GPUs table at developer.nvidia.com/cuda-gpus where RTX PRO 4500 Blackwell is listed with CC 12.0.
  10. Release date of March 18, 2025 per TechPowerUp GPU database. CG Channel reported the announcement on March 18, 2025; OEM availability was slated for summer 2025.
  11. FP4 hardware precision is supported by 5th-generation Tensor Cores. FP8, BF16, TF32, FP16, INT8, and INT4 are also supported by the Blackwell architecture tensor cores. Transformer Engine 2.15+ supports Blackwell (sm_120) for FP8, MXFP8, and NVFP4.
  12. Estimated MSRP of $3,199 USD based on launch-time retail pricing. NVIDIA does not publish an official MSRP for professional workstation GPUs. The PNY retail-boxed variant (VCNRTXPRO4500B-PB) was listed at $3,199.99 on Newegg.com at launch, which is the primary basis for this MSRP. For reference: the NVIDIA-branded OEM variant (900-5G147-2250-000) was listed at $3,018.99 on Newegg.com (different product variant, lower price), and the same PNY SKU was listed at $3,799.00 on B&H Photo (product ID 1898512-REG; higher than Newegg at launch). The $3,199 figure reflects the PNY retail-boxed launch price from Newegg.
  13. NVPN 900-5G147-2250-000 sourced from Newegg product listing (N82E16814132110) for the NVIDIA-branded OEM variant. NVPN 900-5G147-2550-000 sourced from Insight.com product page (insight.com/en_US/shop/product/900-5G147-2550-000). The board_id prefix 5G147 is shared with the RTX PRO 4000 Blackwell (both use the GB203 die), consistent with NVIDIA workstation card naming conventions.
  14. PNY SKU VCNRTXPRO4500B-PB confirmed from Newegg product listing (N82E16814985048). Dell part number 490-BLPL confirmed from dell.com official product page (multiple regional URLs). HP part number B11F2AA confirmed from hp.com GB product page and multiple reseller listings (redcorp.com, provantage.com, gpus.net).