Can Hy Memory 30B A3B run on the L40S?
Yes. At 8K context the best fit is Q8_0, using about 31.1 GB of the 45.6 GB usable on the L40S. Estimated speed ceiling: ~271 tokens/sec.
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 Fits | 23.1 GB Fits | 768 |
| 3-bit (Q3_K_M) | 14.7 GB Fits | 15.1 GB Fits | 17.3 GB Fits | 26.3 GB Fits | 589 |
| 4-bit (Q4_K_M) | 18.1 GB Fits | 18.5 GB Fits | 20.7 GB Fits | 29.7 GB Fits | 471 |
| 5-bit (Q5_K_M) | 21.0 GB Fits | 21.3 GB Fits | 23.6 GB Fits | 32.6 GB Fits | 404 |
| 6-bit (Q6_K) | 24.0 GB Fits | 24.3 GB Fits | 26.6 GB Fits | 35.6 GB Fits | 351 |
| 8-bit (Q8_0) | 30.8 GB Fits | 31.1 GB Fits | 33.4 GB Fits | 42.4 GB Tight | 271 |
| 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 (864 GB/s), not a benchmark. Methodology.
More
Hy Memory 30B A3B on every GPU and Mac · Every model that runs on the L40S · Other models that fit 48 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (tencent/Hy-Memory-30B-A3B), NVIDIA specs. Data as of 2026-11-17.