Can Qwen3.5 27B run on the L40S?
Yes. At 8K context the best fit is Q8_0, using about 28.0 GB of the 45.6 GB usable on the L40S. Estimated speed ceiling: ~30 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) | 10.8 GB Fits | 11.1 GB Fits | 12.6 GB Fits | 18.6 GB Fits | 83 |
| 3-bit (Q3_K_M) • | 13.7 GB Fits | 14.0 GB Fits | 15.5 GB Fits | 21.5 GB Fits | 64 |
| 4-bit (Q4_K_M) • | 16.7 GB Fits | 17.0 GB Fits | 18.5 GB Fits | 24.5 GB Fits | 52 |
| 5-bit (Q5_K_M) • | 19.4 GB Fits | 19.6 GB Fits | 21.1 GB Fits | 27.1 GB Fits | 44 |
| 6-bit (Q6_K) • | 22.0 GB Fits | 22.3 GB Fits | 23.8 GB Fits | 29.8 GB Fits | 38 |
| 8-bit (Q8_0) • | 27.8 GB Fits | 28.0 GB Fits | 29.5 GB Fits | 35.5 GB Fits | 30 |
| 16-bit (F16/BF16) • | 51.2 GB No | 51.5 GB No | 53.0 GB No | 59.0 GB No | — |
• real GGUF file size from unsloth/Qwen3.5-27B-GGUF. Speed is a bandwidth ceiling (864 GB/s), not a benchmark. Methodology.
More
Qwen3.5 27B 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 (Qwen/Qwen3.5-27B), NVIDIA specs. Data as of 2026-11-17.