Qwen3 30B-A3B: how much VRAM?
In Q4_K_M with 8k tokens of context, Qwen3 30B-A3B needs about 20 GB of memory. The cheapest option: 2 × RTX 3060 12 GB (€1,460 for the complete machine).
Model sheet
ollama run qwen3:30bWeights on Hugging Face: Qwen/Qwen3-30B-A3B
VRAM by precision and context
| Precision | 4k | 8k | 32k | 128k |
|---|---|---|---|---|
| FP16 | 65 GB | 65 GB | 67 GB | 76 GB |
| FP8 | 33 GB | 33 GB | 35 GB | 44 GB |
| Q8_0 | 35 GB | 35 GB | 37 GB | 46 GB |
| Q6_K | 27 GB | 27 GB | 30 GB | 39 GB |
| Q4_K_M | 20 GB | 20 GB | 23 GB | 32 GB |
| Q3_K_M | 16 GB | 17 GB | 19 GB | 28 GB |
One user. Each simultaneous user adds their own context cache.
Frequently asked questions
How much VRAM does Qwen3 30B-A3B need?
With an 8k-token context and one user, Qwen3 30B-A3B needs about 20 GB in Q4, 35 GB in Q8 and 65 GB in FP16.
What is the cheapest hardware to run Qwen3 30B-A3B?
In Q4, the cheapest option in our catalogue to run Qwen3 30B-A3B is 2 × RTX 3060 12 GB, about €1,460 for the complete machine.
How fast does Qwen3 30B-A3B run?
Expect around 139 tokens per second for one user on 2 × RTX 3060 12 GB in Q4 (estimate based on memory bandwidth).
On which hardware?
| Hardware | Cards | Speed | Machine |
|---|---|---|---|
| RTX 3060 12 GB | 2 cards | ~139 tok/s | €1,460 |
| RTX 3090 24 GB (used) | 1 is enough | ~242 tok/s | €1,650 |
| RTX 4060 Ti 16 GB | 2 cards | ~112 tok/s | €1,800 |
| Radeon RX 7900 XTX 24 GB | 1 is enough | ~248 tok/s | €1,850 |
| Radeon RX 7800 XT 16 GB | 2 cards | ~242 tok/s | €1,900 |
| RTX 4070 12 GB | 2 cards | ~195 tok/s | €2,000 |
| Mini-PC Ryzen AI Max+ 395 128 GB | 1 is enough | ~66 tok/s | €2,300 |
| RTX 4070 Ti SUPER 16 GB | 2 cards | ~260 tok/s | €2,600 |
| RTX 4090 24 GB | 1 is enough | ~260 tok/s | €2,900 |
| RTX 4080 SUPER 16 GB | 2 cards | ~285 tok/s | €3,000 |
| RTX 5080 16 GB | 2 cards | ~372 tok/s | €3,300 |
| RTX 5090 32 GB | 1 is enough | ~463 tok/s | €3,500 |
| NVIDIA DGX Spark 128 GB | 1 is enough | ~70 tok/s | €4,000 |
| Mac Studio M4 Max 128 GB | 1 is enough | ~141 tok/s | €4,400 |
| RTX 6000 Ada 48 GB | 1 is enough | ~248 tok/s | €7,900 |
| RTX PRO 6000 Blackwell 96 GB | 1 is enough | ~463 tok/s | €9,900 |
| Mac Studio M3 Ultra 512 GB | 1 is enough | ~211 tok/s | €11,500 |
| L40S 48 GB | 1 is enough | ~223 tok/s | €13,500 |
| A100 80 GB | 1 is enough | ~527 tok/s | €17,000 |
| H100 80 GB | 1 is enough | ~865 tok/s | €34,000 |
| H200 141 GB | 1 is enough | ~1,239 tok/s | €39,000 |
| B200 192 GB | 1 is enough | ~2,066 tok/s | €48,000 |
Estimated complete machine: cards + host PC (€900 for 1–2 consumer cards, €2,800 beyond, a €6,000 server for datacenter cards). Macs and mini-PCs are sold complete.
Similar models
From: RTX 3060 12 GB · €1,180
From: RTX 3060 12 GB · €1,180
From: RTX 3060 12 GB · €1,180
From: RTX 4060 Ti 16 GB · €1,350