Qwen3 4B: how much VRAM?
In Q4_K_M with 8k tokens of context, Qwen3 4B needs about 4 GB of memory. The cheapest option: RTX 3060 12 GB (€1,180 for the complete machine).
Model sheet
ollama run qwen3:4bWeights on Hugging Face: Qwen/Qwen3-4B
VRAM by precision and context
| Precision | 4k | 8k | 32k | 128k |
|---|---|---|---|---|
| FP16 | 9 GB | 10 GB | 13 GB | 27 GB |
| FP8 | 5 GB | 6 GB | 9 GB | 23 GB |
| Q8_0 | 5 GB | 6 GB | 9 GB | 23 GB |
| Q6_K | 4 GB | 5 GB | 8 GB | 22 GB |
| Q4_K_M | 3 GB | 4 GB | 7 GB | 21 GB |
| Q3_K_M | 3 GB | 4 GB | 7 GB | 20 GB |
One user. Each simultaneous user adds their own context cache.
Frequently asked questions
How much VRAM does Qwen3 4B need?
With an 8k-token context and one user, Qwen3 4B needs about 4 GB in Q4, 6 GB in Q8 and 10 GB in FP16.
What is the cheapest hardware to run Qwen3 4B?
In Q4, the cheapest option in our catalogue to run Qwen3 4B is RTX 3060 12 GB, about €1,180 for the complete machine.
How fast does Qwen3 4B run?
Expect around 78 tokens per second for one user on RTX 3060 12 GB in Q4 (estimate based on memory bandwidth).
On which hardware?
| Hardware | Cards | Speed | Machine |
|---|---|---|---|
| RTX 3060 12 GB | 1 is enough | ~78 tok/s | €1,180 |
| RTX 4060 Ti 16 GB | 1 is enough | ~62 tok/s | €1,350 |
| Radeon RX 7800 XT 16 GB | 1 is enough | ~135 tok/s | €1,400 |
| RTX 4070 12 GB | 1 is enough | ~109 tok/s | €1,450 |
| RTX 3090 24 GB (used) | 1 is enough | ~202 tok/s | €1,650 |
| RTX 4070 Ti SUPER 16 GB | 1 is enough | ~145 tok/s | €1,750 |
| Radeon RX 7900 XTX 24 GB | 1 is enough | ~207 tok/s | €1,850 |
| RTX 4080 SUPER 16 GB | 1 is enough | ~159 tok/s | €1,950 |
| RTX 5080 16 GB | 1 is enough | ~207 tok/s | €2,100 |
| Mini-PC Ryzen AI Max+ 395 128 GB | 1 is enough | ~55 tok/s | €2,300 |
| RTX 4090 24 GB | 1 is enough | ~217 tok/s | €2,900 |
| RTX 5090 32 GB | 1 is enough | ~387 tok/s | €3,500 |
| NVIDIA DGX Spark 128 GB | 1 is enough | ~59 tok/s | €4,000 |
| Mac Studio M4 Max 128 GB | 1 is enough | ~118 tok/s | €4,400 |
| RTX 6000 Ada 48 GB | 1 is enough | ~207 tok/s | €7,900 |
| RTX PRO 6000 Blackwell 96 GB | 1 is enough | ~387 tok/s | €9,900 |
| Mac Studio M3 Ultra 512 GB | 1 is enough | ~177 tok/s | €11,500 |
| L40S 48 GB | 1 is enough | ~186 tok/s | €13,500 |
| A100 80 GB | 1 is enough | ~440 tok/s | €17,000 |
| H100 80 GB | 1 is enough | ~723 tok/s | €34,000 |
| H200 141 GB | 1 is enough | ~1,035 tok/s | €39,000 |
| B200 192 GB | 1 is enough | ~1,725 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: 2 × RTX 3060 12 GB · €1,460
From: 2 × RTX 3060 12 GB · €1,460