Can I run Hy-Memory-30B-A3B on an RTX 4090?

Yes, with a GGUF version

The published weights are too big, but the Q4_K_M file from tencent/Hy-Memory-30B-A3B-GGUF (18.2 GB) fits, with room for ≈ 33K tokens of context.

Full analysis of Hy-Memory-30B-A3B Everything that runs on RTX 4090 Get the Q4_K_M file on Hugging Face

ollama run hf.co/tencent/Hy-Memory-30B-A3B-GGUF:Q4_K_M

Files and context on RTX 4090

Each file of Hy-Memory-30B-A3B (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 RTX 4090	Context (16-bit)	Context (8-bit)
As published	60.1 GB	Doesn't fit	—	—
Hy-Memory-30B-A3B-Q8_0tencent/Hy-Memory-30B-A3B-GGUF	32.0 GB	Doesn't fit	—	—
Q6_Ktencent/Hy-Memory-30B-A3B-GGUF	24.7 GB	Doesn't fit	—	—
Q4_K_Mtencent/Hy-Memory-30B-A3B-GGUF	18.2 GB	Tight	33K	67K
Q3_K_Mtencent/Hy-Memory-30B-A3B-GGUF	14.4 GB	Fits	71K	142K
Hy-Memory-30B-A3B-Q8_0-expsIQ2_XXStencent/Hy-Memory-30B-A3B-GGUF	9.14 GB	Fits	124K	247K
Q8_0tencent/Hy-Memory-30B-A3B-GGUF	8.03 GB	Fits	135K	256K

Hy-Memory-30B-A3B on other hardware

Popular models that run on RTX 4090

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.