RunLocalAI
amd
APU
128GB VRAM
workstation

AMD Instinct MI300A (APU)

Combined CPU + GPU APU with 128GB unified HBM3. Powers the El Capitan supercomputer.

Released 2023

Overview

Combined CPU + GPU APU with 128GB unified HBM3. Powers the El Capitan supercomputer.

Specs

VRAM128 GB
Power draw760 W
Released2023
Backends
ROCm

Models that fit

Open-weight models small enough to run on AMD Instinct MI300A (APU) with usable context.

Llama 3.1 8B Instruct
8B · llama
Qwen 3 30B-A3B
30B · qwen
Qwen 2.5 Coder 32B Instruct
32B · qwen
Llama 3.3 70B Instruct
70B · llama
Qwen 3 32B
32B · qwen
Gemma 4 31B Dense
31B · gemma
Qwen 3 8B
8B · qwen
DeepSeek R1 Distill Llama 70B
70B · deepseek

Frequently asked

What models can AMD Instinct MI300A (APU) run?

With 128GB VRAM, the AMD Instinct MI300A (APU) runs 70B models in 4-bit quantization, plus everything smaller. See the model list below for tested combinations.

Does AMD Instinct MI300A (APU) support CUDA?

No — AMD Instinct MI300A (APU) is an AMD card. Use ROCm (Linux) or the Vulkan backend in llama.cpp instead. CUDA-only tools won't work.

Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify hardware specifications.