NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition

Power-capped "Max-Q" variant of the RTX PRO 6000 Blackwell Workstation Edition — a distinct NVIDIA SKU with its own datasheet, not the same part as nvidia-rtx-pro-6000-blackwell-96gb (600W). 300W total board power (vs. 600W for the standard Workstation Edition), active dual-slot cooling, 24,064 CUDA cores, 5th-gen Tensor Cores (FP4 support), 4th-gen RT Cores, 1792 GB/s memory bandwidth, PCIe 5.0 x16. NVIDIA's datasheet publishes "3511 AI TOPS" footnoted as an effective FP4-with-sparsity figure (not FP8) — recorded here as fp8Tflops for lack of a dedicated FP4 field, per this dataset's convention of using fp8Tflops for the datasheet's headline Tensor/AI performance number. Also publishes 110 TFLOPS single-precision (FP32) and 333 TFLOPS RT Core performance, neither of which has a dedicated schema field.

VRAM

96 GB

Memory

GDDR7

FP16

—

FP8

3511 TF