Can I run PowerMoE-3b on a Radeon RX 9070 XT?

Yes, just

Its published weights (13.5 GB) fit in the 14.4 GB a Radeon RX 9070 XT can use, with room for ≈ 4K tokens of context.

Full analysis of PowerMoE-3b Everything that runs on Radeon RX 9070 XT Get the weights on Hugging Face

hf download ibm-research/PowerMoE-3b

Files and context on Radeon RX 9070 XT

Each file of PowerMoE-3b (and of GGUF versions the site has analyzed), and how much context fits next to it, with the context cache in 16-bit or 8-bit.

File	Size	On Radeon RX 9070 XT	Context (16-bit)	Context (8-bit)
As published	13.5 GB	Tight	4K	4K
4-bit (estimate)	1.90 GB	Fits	4K	4K

PowerMoE-3b on other hardware

NVIDIA H100Yes
RTX 5090Yes
RTX 4090Yes
RTX 3090Yes
RTX 5060 Ti 16 GBYes, just
RTX 3060 12 GBProbably, at 4-bit
RTX 4060Probably, at 4-bit
Radeon RX 9070 XTYes, just
Mac with 32 GBYes
Mac with 64 GBYes
Mac with 128 GBYes

Popular models that run on Radeon RX 9070 XT

Qwen3-0.6BQwen · 752M
gpt2openai-community · 137M
Qwen3-Coder-30B-A3B-Instruct-GGUFunsloth · 30.5B
Qwen2.5-0.5B-InstructQwen · 494M
Qwen2.5-1.5B-InstructQwen · 1.54B
Qwen3-4BQwen · 4.02B
Qwen3.5-4BQwen · 4.66B
gpt-oss-20bopenai · 20.9B

Assumes 90% of the card's memory is usable and at least 4K tokens of context for one conversation, with the context cache in 16-bit unless noted. Fits leaves at least 20% of that memory for context; Tight leaves less.