Can Hy Memory 30B A3B run on the RTX 5060 Ti 16GB?
Yes. At 8K context the best fit is Q2_K, using about 11.9 GB of the 15.2 GB usable on the RTX 5060 Ti 16GB. Estimated speed ceiling: ~398 tokens/sec.
Better quality with offload: 4-bit (Q4_K_M)
Runs with offload 4-bit (Q4_K_M) at 8K context: all attention layers plus the experts of 36 of 47 MoE layers on the RTX 5060 Ti 16GB (about 13.9 GB), and the experts of the other 11 layers in system RAM (about 3.9 GB). Needs a PC with at least 32 GB of system RAM. Estimated speed: 65–106 tokens/sec (usable for chat), versus a bandwidth ceiling of ~244 tok/s if it fit fully in VRAM.
llama-server -hf tencent/Hy-Memory-30B-A3B-GGUF:Q4_K_M -c 8192 -ngl 999 --n-cpu-moe 12
Estimate, not a benchmark. The command downloads the GGUF from Hugging Face. The speed range spans dual-channel DDR4-3200 to DDR5-6000 system RAM; your CPU, RAM speed and llama.cpp version move it. If llama.cpp runs out of GPU memory, raise --n-cpu-moe by 1-2; if VRAM is left over, lower it for more speed. Recent llama.cpp builds can also pick the split automatically. How we estimate offload.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | 128K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|---|
| 2-bit (Q2_K) | 11.5 GB Fits | 11.9 GB Fits | 14.1 GB Tight | 23.1 GB No | 398 |
| 3-bit (Q3_K_M) | 14.7 GB Tight | 15.1 GB Tight | 17.3 GB No | 26.3 GB No | 306 |
| 4-bit (Q4_K_M) | 18.1 GB No | 18.5 GB No | 20.7 GB No | 29.7 GB No | — |
| 5-bit (Q5_K_M) | 21.0 GB No | 21.3 GB No | 23.6 GB No | 32.6 GB No | — |
| 6-bit (Q6_K) | 24.0 GB No | 24.3 GB No | 26.6 GB No | 35.6 GB No | — |
| 8-bit (Q8_0) | 30.8 GB No | 31.1 GB No | 33.4 GB No | 42.4 GB No | — |
| 16-bit (F16/BF16) | 57.0 GB No | 57.4 GB No | 59.6 GB No | 68.6 GB No | — |
• real GGUF file size. Speed is a bandwidth ceiling (448 GB/s), not a benchmark. Methodology.
What fits instead
- Cheapest listed GPU that runs Hy Memory 30B A3B: RX 7900 XTX (24 GB, launch MSRP $999) at Q4_K_M, fully in VRAM at 8K context.
- Largest popular model that fits the RTX 5060 Ti 16GB: gpt-oss 20B (21B, MoE) at Q8_0, about 12.3 GB at 8K context.
"Runs" and "fits" mean the whole model in GPU memory at 8K context (4-bit or better for the model pick). Prices are US launch MSRPs, not today's street prices. "Popular" means the 60 most-downloaded models we track.
More
Hy Memory 30B A3B on every GPU and Mac · Every model that runs on the RTX 5060 Ti 16GB · Other models that fit 16 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (tencent/Hy-Memory-30B-A3B), NVIDIA specs. Model data last checked 2026-11-17; refreshed hourly.