Specifications
Vendor
AMD
Model
Radeon AI PRO R9700
Architecture
RDNA 4 (Navi 48)
Compute Units
64 (4,096 stream processors)
AI Accelerators
128 (2nd generation)
Memory
32 GB GDDR6, 256-bit
Memory Bandwidth
640 GB/s
Infinity Cache
64 MB
Interface
PCIe 5.0 x16
Peak FP16
95.7 TFLOPS
Peak INT4
1,531 TOPS
Boost Clock
up to 2,920 MHz
Total Board Power
300 W
Cooling
Active (2-slot)
Memory ECC
Yes (Linux)
Applications
Local AI inference, LLM fine-tuning, CAD, rendering, VDI
Overview
The AMD Radeon AI PRO R9700 is a workstation-class graphics card built for local AI work. Based on the RDNA 4 (Navi 48) architecture, it carries 32 GB of GDDR6 memory on a 256-bit bus, giving 640 GB/s of bandwidth and 64 MB of Infinity Cache - enough to hold medium-to-large models on the card itself.
Its 128 second-generation AI accelerators deliver up to 1,531 TOPS of INT4 and 95.7 TFLOPS of FP16 performance, while a 300 W board power and 2-slot active cooler keep it deployable in standard workstation chassis. ECC memory (Linux), PCIe 5.0 x16 and broad support across ML frameworks make it suitable for inference servers, LLM stations and AI research workstations.
Key Benefits
32 GB GDDR6 · RDNA 4 (Navi 48) · 128 AI accelerators · 640 GB/s · up to 1,531 TOPS INT4 · PCIe 5.0 x16 · ECC (Linux).
Applications
Local LLM inference and fine-tuning workstations, AI development and research, CAD/CAE, media and rendering, and VDI/remote-workstation deployments needing 32 GB of GPU memory.
Request a Quote — AMD Radeon AI PRO R9700 32GB
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →