Specifications
Model
G4L3-ZX1-LAT4
Series
GIGABYTE G4L3 AMD HPC/AI GPU server
Rack Unit Size
4U
Processor
Dual AMD EPYC 9005 / 9004 Series
GPUs
8x AMD Instinct MI355X
Cooling
Direct liquid cooling (DLC)
Vendor
GIGABYTE
Target Workloads
AI training, AI inference, HPC
Related Series Models
G494-SB0-LAP1, G294-Z22-AAP2, G4L4-SD3-LAX7
Overview
The GIGABYTE G4L3-ZX1-LAT4 is the AMD-accelerated tier of GIGABYTE's 4U HPC/AI server line: eight AMD Instinct MI355X GPUs paired with dual AMD EPYC 9005 or 9004 series processors. It is the direct alternative to the Intel-based G4L4-SD3-LAX7 for buyers standardising on an all-AMD compute stack.
Sourcing GPU and CPU from the same vendor removes the mixed-fabric tuning work that comes with an AMD host and an NVIDIA accelerator baseboard, and it keeps host-to-device PCIe topology inside one validation envelope.
GIGABYTE ships the chassis with direct liquid cooling, consistent with the rest of its MI355X and Blackwell Ultra 4U platforms, so the node can share a CDU loop with GB300-class racks in the same hall.
Key Benefits
Single-vendor compute stack: EPYC host and Instinct accelerators share one validation and driver path. DLC by default: the 4U chassis is plumbed for direct liquid cooling so it slots into an existing CDU loop. HPC and AI on one node: MI355X density with dual EPYC host cores covers mixed simulation and training fleets.
Applications
AI training and inference on AMD Instinct accelerators, HPC simulation, scientific computing clusters, sovereign AI builds requiring a non-NVIDIA accelerator path, and mixed HPC/AI data-centre deployments.
Request a Quote — GIGABYTE G4L3-ZX1-LAT4 — 4U Dual EPYC Server with 8x AMD Instinct MI355X
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →