What GPU do you need to run Qwen3.5 9B?
9B params 256K max context VisionTool useReasoning
The smallest GPU memory that runs Qwen3.5 9B well is 6 GB (Q4_K_M, e.g. GeForce RTX 3050 6GB). 128 of 132 GPUs run it well.
MMLU 82.5
VRAM needed per quantization
| Quantization | Weights | 16K context | 32K context | Full 256K |
|---|---|---|---|---|
| Q4_K_M | 5.3 GB | 5.8 GB | 6.3 GB | 13.2 GB |
| Q8_0 | 8.9 GB | 9.4 GB | 9.9 GB | 16.8 GB |
Weights plus fp16 KV cache. Leave ~1 GB headroom for the inference engine.
Cheapest GPUs that can run Qwen3.5 9B
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 Qwen3.5 9B
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
- GTX 1060 6GB 6 GB Q4_K_M 22K context · ~31 tok/s
- GTX 1660 6 GB Q4_K_M 22K context · ~31 tok/s
- GTX 1660 SUPER 6 GB Q4_K_M 22K context · ~53 tok/s
- GTX 1660 Ti 6 GB Q4_K_M 22K context · ~46 tok/s
- RTX 2060 6 GB Q4_K_M 22K context · ~53 tok/s
- RTX 3050 6GB 6 GB Q4_K_M 22K context · ~27 tok/s
- RTX A2000 6 GB Q4_K_M 22K context · ~46 tok/s
- GTX 1070 8 GB Q4_K_M 87K context · ~41 tok/s
- GTX 1070 Ti 8 GB Q4_K_M 87K context · ~41 tok/s
- GTX 1080 8 GB Q4_K_M 87K context · ~51 tok/s
- RTX 2060 SUPER 8 GB Q4_K_M 87K context · ~71 tok/s
- RTX 2070 8 GB Q4_K_M 87K context · ~71 tok/s
- RTX 2070 SUPER 8 GB Q4_K_M 87K context · ~71 tok/s
- RTX 2080 8 GB Q4_K_M 87K context · ~71 tok/s
- RTX 2080 SUPER 8 GB Q4_K_M 87K context · ~79 tok/s
- RTX 3050 8 GB Q4_K_M 87K context · ~36 tok/s
- RTX 3060 Ti 8 GB Q4_K_M 87K context · ~71 tok/s
- RTX 3060 Ti GDDR6X 8 GB Q4_K_M 87K context · ~97 tok/s
- RTX 3070 8 GB Q4_K_M 87K context · ~71 tok/s
- RTX 3070 Ti 8 GB Q4_K_M 87K context · ~97 tok/s
- RTX 4060 8 GB Q4_K_M 87K context · ~43 tok/s
- RTX 4060 Ti 8 GB Q4_K_M 87K context · ~46 tok/s
- RTX 5050 8 GB Q4_K_M 87K context · ~51 tok/s
- RTX 5060 8 GB Q4_K_M 87K context · ~71 tok/s
- RTX 5060 Ti 8 GB Q4_K_M 87K context · ~71 tok/s
- T1000 8GB 8 GB Q4_K_M 87K context · ~25 tok/s
- RTX 3080 10 GB Q8_0 36K context · ~77 tok/s
- GTX 1080 Ti 11 GB Q8_0 68K context · ~49 tok/s
- RTX 2080 Ti 11 GB Q8_0 68K context · ~62 tok/s
- RTX 2060 12GB 12 GB Q8_0 100K context · ~34 tok/s
- RTX 3060 12 GB Q8_0 100K context · ~36 tok/s
- RTX 3080 12GB 12 GB Q8_0 100K context · ~92 tok/s
- RTX 3080 Ti 12 GB Q8_0 100K context · ~92 tok/s
- RTX 4070 12 GB Q8_0 100K context · ~51 tok/s
- RTX 4070 SUPER 12 GB Q8_0 100K context · ~51 tok/s
- RTX 4070 Ti 12 GB Q8_0 100K context · ~51 tok/s
- RTX 5070 12 GB Q8_0 100K context · ~68 tok/s
- RTX A2000 12GB 12 GB Q8_0 100K context · ~29 tok/s
- RTX 4060 Ti 16GB 16 GB Q8_0 230K context · ~29 tok/s
- RTX 4070 Ti SUPER 16 GB Q8_0 230K context · ~68 tok/s
- RTX 4080 16 GB Q8_0 230K context · ~73 tok/s
- RTX 4080 SUPER 16 GB Q8_0 230K context · ~75 tok/s
- RTX 5060 Ti 16GB 16 GB Q8_0 230K context · ~45 tok/s
- RTX 5070 Ti 16 GB Q8_0 230K context · ~91 tok/s
- RTX 5080 16 GB Q8_0 230K context · ~97 tok/s
- RTX 2000 Ada 16 GB Q8_0 230K context · ~23 tok/s
- RTX A4000 16 GB Q8_0 230K context · ~45 tok/s
- RTX PRO 2000 Blackwell 16 GB Q8_0 230K context · ~29 tok/s
- Tesla T4 16 GB Q8_0 230K context · ~32 tok/s
- Tesla V100 16GB 16 GB Q8_0 230K context · ~91 tok/s
- RTX 4000 Ada 20 GB Q8_0 256K context · ~36 tok/s
- RTX 3090 24 GB Q8_0 256K context · ~95 tok/s
- RTX 3090 Ti 24 GB Q8_0 256K context · ~102 tok/s
- RTX 4090 24 GB Q8_0 256K context · ~102 tok/s
- RTX 4500 Ada 24 GB Q8_0 256K context · ~44 tok/s
- RTX A5000 24 GB Q8_0 256K context · ~78 tok/s
- RTX PRO 4000 Blackwell 24 GB Q8_0 256K context · ~68 tok/s
- RTX Spark 24 GB Q8_0 256K context · ~30 tok/s
- Tesla P40 24 GB Q8_0 256K context · ~35 tok/s
- RTX 5090 32 GB Q8_0 256K context · ~182 tok/s
- RTX 5000 Ada 32 GB Q8_0 256K context · ~58 tok/s
- RTX PRO 4500 Blackwell 32 GB Q8_0 256K context · ~91 tok/s
- Tesla V100 32GB 32 GB Q8_0 256K context · ~91 tok/s
- L40S 48 GB Q8_0 256K context · ~88 tok/s
- RTX 6000 Ada 48 GB Q8_0 256K context · ~97 tok/s
- RTX A6000 48 GB Q8_0 256K context · ~78 tok/s
- RTX PRO 5000 Blackwell 48 GB Q8_0 256K context · ~136 tok/s
- RTX PRO 5000 Blackwell 72GB 72 GB Q8_0 256K context · ~136 tok/s
- RTX PRO 5500 Blackwell 84 GB Q8_0 256K context · ~142 tok/s
- RTX PRO 6000 Blackwell 96 GB Q8_0 256K context · ~182 tok/s
- DGX Spark (GB 10) 128 GB Q8_0 256K context · ~28 tok/s
AMD
- Radeon RX 5600 XT 6 GB Q4_K_M 22K context · ~46 tok/s
- Radeon RX 5700 8 GB Q4_K_M 87K context · ~71 tok/s
- Radeon RX 5700 XT 8 GB Q4_K_M 87K context · ~71 tok/s
- Radeon RX 6600 8 GB Q4_K_M 87K context · ~36 tok/s
- Radeon RX 6600 XT 8 GB Q4_K_M 87K context · ~41 tok/s
- Radeon RX 6650 XT 8 GB Q4_K_M 87K context · ~45 tok/s
- Radeon RX 7600 8 GB Q4_K_M 87K context · ~46 tok/s
- Radeon RX 9060 8 GB Q4_K_M 87K context · ~46 tok/s
- Radeon RX 9060 XT 8 GB Q4_K_M 87K context · ~51 tok/s
- Radeon RX 6700 XT 12 GB Q8_0 100K context · ~39 tok/s
- Radeon RX 6750 XT 12 GB Q8_0 100K context · ~44 tok/s
- Radeon RX 7700 XT 12 GB Q8_0 100K context · ~44 tok/s
- Radeon RX 9070 GRE 12 GB Q8_0 100K context · ~44 tok/s
- AMD BC-250 16 GB Q8_0 230K context · ~45 tok/s
- Radeon RX 6800 16 GB Q8_0 230K context · ~52 tok/s
- Radeon RX 6800 XT 16 GB Q8_0 230K context · ~52 tok/s
- Radeon RX 6900 XT 16 GB Q8_0 230K context · ~52 tok/s
- Radeon RX 6950 XT 16 GB Q8_0 230K context · ~58 tok/s
- Radeon RX 7600 XT 16 GB Q8_0 230K context · ~29 tok/s
- Radeon RX 7800 XT 16 GB Q8_0 230K context · ~63 tok/s
- Radeon RX 7900 GRE 16 GB Q8_0 230K context · ~58 tok/s
- Radeon RX 9060 XT 16GB 16 GB Q8_0 230K context · ~32 tok/s
- Radeon RX 9070 16 GB Q8_0 230K context · ~65 tok/s
- Radeon RX 9070 XT 16 GB Q8_0 230K context · ~65 tok/s
- Radeon RX 7900 XT 20 GB Q8_0 256K context · ~81 tok/s
- Radeon RX 7900 XTX 24 GB Q8_0 256K context · ~97 tok/s
- Radeon AI PRO R9700 32 GB Q8_0 256K context · ~65 tok/s
- Ryzen AI Max (Strix Halo) 32 GB Q8_0 256K context · ~26 tok/s
- Ryzen AI Max 400 (Gorgon Halo) 64 GB Q8_0 256K context · ~28 tok/s
Intel
- Arc A580 8 GB Q4_K_M 87K context · ~81 tok/s
- Arc A750 8 GB Q4_K_M 87K context · ~81 tok/s
- Arc A770 8GB 8 GB Q4_K_M 87K context · ~81 tok/s
- Arc B570 10 GB Q8_0 36K context · ~39 tok/s
- Arc B580 12 GB Q8_0 100K context · ~46 tok/s
- Arc A770 16GB 16 GB Q8_0 230K context · ~57 tok/s
- Arc Pro B65 32 GB Q8_0 256K context · ~62 tok/s
- Arc Pro B70 32 GB Q8_0 256K context · ~62 tok/s
Apple
- M1 8 GB Q4_K_M 87K context · ~11 tok/s
- M2 8 GB Q4_K_M 87K context · ~16 tok/s
- M3 8 GB Q4_K_M 87K context · ~16 tok/s
- M1 Pro 16 GB Q8_0 230K context · ~20 tok/s
- M2 Pro 16 GB Q8_0 230K context · ~20 tok/s
- M4 16 GB Q8_0 230K context · ~12 tok/s
- M5 16 GB Q8_0 230K context · ~16 tok/s
- M6 16 GB Q8_0 230K context · ~16 tok/s
- M3 Pro 18 GB Q8_0 256K context · ~15 tok/s
- M4 Pro 24 GB Q8_0 256K context · ~28 tok/s
- M5 Pro 24 GB Q8_0 256K context · ~31 tok/s
- M1 Max 32 GB Q8_0 256K context · ~41 tok/s
- M2 Max 32 GB Q8_0 256K context · ~41 tok/s
- M3 Max 36 GB Q8_0 256K context · ~30 tok/s
- M4 Max 36 GB Q8_0 256K context · ~42 tok/s
- M5 Max 36 GB Q8_0 256K context · ~47 tok/s
- M1 Ultra 64 GB Q8_0 256K context · ~81 tok/s
- M2 Ultra 64 GB Q8_0 256K context · ~81 tok/s
- M3 Ultra 96 GB Q8_0 256K context · ~83 tok/s
- M5 Ultra 96 GB Q8_0 256K context · ~122 tok/s