Announced at GTC 2026 and expected in the second half of 2026. This rack is also the smallest unit available: the Groq 3 LPU is soldered eight to a tray and there is no standalone accelerator SKU.

NVIDIA Groq 3 LPX

rack
256× Groq 3 LPU
LPU C2C (chip-to-chip) / C2C spine
2026
FP8
307
PFLOPS
Power
—
kW Total
Memory
—
GB Total

We will point you at suppliers who have it. Free, and no signup.

FP8
307
PFLOPS

System Details

GPU Configuration

GPU Count: 256 GPUs
Architecture: Groq LPU
Interconnect: LPU C2C (chip-to-chip) / C2C spine

System Specifications

Form Factor: rack
Total Power: —
Total Memory: —
On-die SRAM: 128 GB
Memory Bandwidth: —

Precision Performance Breakdown

PrecisionSystem PerformancePer GPUEfficiency
307 PFLOPS 1,200 TFLOPS —

Where a vendor states a basis, the figure shown is the dense one, and any figure it publishes is listed separately. Vendors commonly headline the sparse number instead: for NVIDIA's 2:4 structured sparsity that is exactly twice the dense figure, though other vendors' sparse modes do not all follow that ratio.

Powered by NVIDIA Groq 3 LPU

This system utilizes 256 × NVIDIA Groq 3 LPU GPUs, each delivering exceptional performance for AI and HPC workloads.

Per GPU TDP

—

Per GPU Memory

—

Process Node

—

Architecture

Groq LPU

Documentation & Resources

Flopper Spec Sheet

NVIDIA Groq 3 LPX specifications, generated from our database. Printable.

Vendor product page

NVIDIA Groq 3 LPX documentation

View ↗

NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition Datasheet

NVIDIA • 2026-06-01

View Document ↗

NVIDIA DGX Documentation

System guides, deployment resources

Browse ↗

Typical Use Cases

AI/ML Training
High-Performance Computing
Data Analytics

The NVIDIA Groq 3 LPX runs 256× Groq 3 LPU GPUs over LPU C2C (chip-to-chip) / C2C spine, delivering 307 PFLOPS FP8 dense.

© 2026 Flopper.io - Compare the GPUs Powering AI