Can Qwen3.8 27B run on the RTX 5070?
Yes. At 8K context the best fit is Q2_K, using about 11.1 GB of the 11.4 GB usable on the RTX 5070. Estimated speed ceiling: ~65 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 Tight | 11.1 GB Tight | 12.6 GB No | 18.6 GB No | 65 |
| 3-bit (Q3_K_M) | 13.8 GB No | 14.0 GB No | 15.5 GB No | 21.5 GB No | — |
| 4-bit (Q4_K_M) • | 16.6 GB No | 16.8 GB No | 18.3 GB No | 24.3 GB No | — |
| 5-bit (Q5_K_M) • | 19.6 GB No | 19.9 GB No | 21.4 GB No | 27.4 GB No | — |
| 6-bit (Q6_K) • | 21.6 GB No | 21.8 GB No | 23.3 GB No | 29.3 GB No | — |
| 8-bit (Q8_0) • | 28.2 GB No | 28.4 GB No | 29.9 GB No | 35.9 GB No | — |
| 16-bit (F16/BF16) • | 52.0 GB No | 52.3 GB No | 53.8 GB No | 59.8 GB No | — |
• real GGUF file size from unsloth/Qwen3.8-27B-GGUF. Speed is a bandwidth ceiling (672 GB/s), not a benchmark. Methodology.
More
Qwen3.8 27B on every GPU and Mac · Every model that runs on the RTX 5070 · Other models that fit 12 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (Qwen/Qwen3.8-27B), NVIDIA specs. Data as of 2026-11-17.