What GPU do you need to run Qwen3.6 27B?
27B params 256K max context VisionTool useReasoning
The smallest GPU memory that runs Qwen3.6 27B well is 18 GB (Q4_K_M, e.g. M3 Pro). 47 of 132 GPUs run it well.
MMLU 86.2 SWE-bench Verified 77.2
VRAM needed per quantization
| Quantization | Weights | 16K context | 32K context | Full 256K |
|---|---|---|---|---|
| Q4_K_M | 16.5 GB | 17.5 GB | 18.5 GB | 32.9 GB |
| Q8_0 | 28.0 GB | 29.0 GB | 30.0 GB | 44.4 GB |
Weights plus fp16 KV cache. Leave ~1 GB headroom for the inference engine.
Cheapest GPUs that can run Qwen3.6 27B
Lowest launch MSRP among GPUs that run its smallest quantization (Q4_K_M) well, with at least 16K context.
As an Amazon Associate, I earn from qualifying purchases.
GPUs that can run Qwen3.6 27B
Best quantization that fits on each GPU, with the context and speed it gets. For chips with several memory sizes, the smallest that works is shown.
NVIDIA
- RTX 4000 Ada 20 GB Q4_K_M 54K context · ~21 tok/s
- RTX 3090 24 GB Q4_K_M 117K context · ~53 tok/s
- RTX 3090 Ti 24 GB Q4_K_M 117K context · ~58 tok/s
- RTX 4090 24 GB Q4_K_M 117K context · ~58 tok/s
- RTX 4500 Ada 24 GB Q4_K_M 117K context · ~25 tok/s
- RTX A5000 24 GB Q4_K_M 117K context · ~44 tok/s
- RTX PRO 4000 Blackwell 24 GB Q4_K_M 117K context · ~38 tok/s
- RTX Spark 24 GB Q4_K_M 117K context · ~17 tok/s
- Tesla P40 24 GB Q4_K_M 117K context · ~20 tok/s
- RTX 5090 32 GB Q8_0 62K context · ~62 tok/s
- RTX 5000 Ada 32 GB Q8_0 62K context · ~20 tok/s
- RTX PRO 4500 Blackwell 32 GB Q8_0 62K context · ~31 tok/s
- Tesla V100 32GB 32 GB Q8_0 62K context · ~31 tok/s
- L40S 48 GB Q8_0 256K context · ~30 tok/s
- RTX 6000 Ada 48 GB Q8_0 256K context · ~33 tok/s
- RTX A6000 48 GB Q8_0 256K context · ~26 tok/s
- RTX PRO 5000 Blackwell 48 GB Q8_0 256K context · ~46 tok/s
- RTX PRO 5000 Blackwell 72GB 72 GB Q8_0 256K context · ~46 tok/s
- RTX PRO 5500 Blackwell 84 GB Q8_0 256K context · ~48 tok/s
- RTX PRO 6000 Blackwell 96 GB Q8_0 256K context · ~62 tok/s
- DGX Spark (GB 10) 128 GB Q8_0 256K context · ~9 tok/s
AMD
- Radeon RX 7900 XT 20 GB Q4_K_M 54K context · ~46 tok/s
- Radeon RX 7900 XTX 24 GB Q4_K_M 117K context · ~55 tok/s
- Radeon AI PRO R9700 32 GB Q8_0 62K context · ~22 tok/s
- Ryzen AI Max (Strix Halo) 32 GB Q8_0 62K context · ~9 tok/s
- Ryzen AI Max 400 (Gorgon Halo) 64 GB Q8_0 256K context · ~9 tok/s
Intel
- Arc Pro B65 32 GB Q8_0 62K context · ~21 tok/s
- Arc Pro B70 32 GB Q8_0 62K context · ~21 tok/s
Apple
- M3 Pro 18 GB Q4_K_M 23K context · ~9 tok/s
- M2 24 GB Q4_K_M 117K context · ~6 tok/s
- M3 24 GB Q4_K_M 117K context · ~6 tok/s
- M4 24 GB Q4_K_M 117K context · ~7 tok/s
- M4 Pro 24 GB Q4_K_M 117K context · ~16 tok/s
- M5 24 GB Q4_K_M 117K context · ~9 tok/s
- M5 Pro 24 GB Q4_K_M 117K context · ~18 tok/s
- M6 24 GB Q4_K_M 117K context · ~10 tok/s
- M1 Max 32 GB Q8_0 62K context · ~14 tok/s
- M1 Pro 32 GB Q8_0 62K context · ~7 tok/s
- M2 Max 32 GB Q8_0 62K context · ~14 tok/s
- M2 Pro 32 GB Q8_0 62K context · ~7 tok/s
- M3 Max 36 GB Q8_0 125K context · ~10 tok/s
- M4 Max 36 GB Q8_0 125K context · ~14 tok/s
- M5 Max 36 GB Q8_0 125K context · ~16 tok/s
- M1 Ultra 64 GB Q8_0 256K context · ~28 tok/s
- M2 Ultra 64 GB Q8_0 256K context · ~28 tok/s
- M3 Ultra 96 GB Q8_0 256K context · ~28 tok/s
- M5 Ultra 96 GB Q8_0 256K context · ~41 tok/s