Taalai's AI Hardware Co.
Spec sheet · SKU GPU-RTXPRO6000-96G · verified July 2026
RTX PRO 6000 Blackwell
Workstation card In stock

RTX PRO 6000 Blackwell

A professional workstation card built for serious AI work — three times the memory of a desktop card in the same single slot.

SKU GPU-RTXPRO6000-96G
PriceiChecked against NVIDIA's own product listings and major retailers (B&H, Micro Center, NVIDIA's marketplace) as of July 2026 — real published prices, not estimates. $8,565
Request a quote Back to catalog

Specifications

GPU memory
96 GB GDDR7iA fast memory type used on desktop and workstation graphics cards. Cheaper per gigabyte than the HBM used in datacenter GPUs, but with less total bandwidth.
Power draw
600 W
Tensor / AI compute
125 TFLOPS FP32
Rack units
N/A — workstation PCIe card, not rack-mounted
CUDA cores24,064
Memory bandwidth1,792 GB/s
FP32 compute125 TFLOPS
Form factorWorkstation PCIe card

Which models this can run

Real memory math against the 3 models in the model advisor — parameters × 1 GB, plus 20% working room.

Qwen3.5-397B 476.4 GB required · this unit has 96 GB
Does not have enough memory, and this product does not cluster.
DeepSeek-V3.2 805.2 GB required · this unit has 96 GB
Does not have enough memory, and this product does not cluster.
Kimi K2.6 1,200.0 GB required · this unit has 96 GB
Does not have enough memory, and this product does not cluster.

Power and cooling

Electrical requirement: Draws 600 W — fits within a single standard 120V/20A circuit (1,920 W safe continuous per NEC's 80% rule). No special electrical work needed.
About 0.5x the continuous draw of an average home (1200 W). Run flat-out for a month, that's roughly 432 kWh — about $56/month at a $0.13/kWh commercial rate (US average, EIA — your actual rate depends on region and utility contract).
Power cost, relative to the rest of the catalog $56/mo

Cooling: Air-cooled (blower-style, designed for workstation chassis airflow)

What this is good for

  • Running mid-size models (up to roughly 80 GB after overhead) on a single card
  • Workstation-class inference without a full server
  • Teams that need more memory than a desktop card but not a full server

What this is NOT good for

  • Training frontier-scale models
  • High-concurrency production serving
  • Multi-GPU scaling — this card does not cluster with others

Included components

GPU card only. No workstation chassis, PSU, or motherboard included.

Lead time and warranty

Delivery window estimateTypically ships within 3–10 business days from authorized retailers, subject to stock.
WarrantyManufacturer warranty applies (terms vary by reseller).
Request a quote for this system
Get matched to a build