Qwen3 8B: how much VRAM?
In Q4_K_M with 8k tokens of context, Qwen3 8B needs about 7 GB of memory. The cheapest option: RTX 3060 12 GB (€1,180 for the complete machine).
Model sheet
ollama run qwen3:8bWeights on Hugging Face: Qwen/Qwen3-8B
VRAM by precision and context
| Precision | 4k | 8k | 32k | 128k |
|---|---|---|---|---|
| FP16 | 18 GB | 19 GB | 22 GB | 36 GB |
| FP8 | 10 GB | 10 GB | 13 GB | 27 GB |
| Q8_0 | 10 GB | 11 GB | 14 GB | 27 GB |
| Q6_K | 8 GB | 9 GB | 12 GB | 25 GB |
| Q4_K_M | 6 GB | 7 GB | 10 GB | 23 GB |
| Q3_K_M | 5 GB | 6 GB | 9 GB | 23 GB |
One user. Each simultaneous user adds their own context cache.
Frequently asked questions
How much VRAM does Qwen3 8B need?
With an 8k-token context and one user, Qwen3 8B needs about 7 GB in Q4, 11 GB in Q8 and 19 GB in FP16.
What is the cheapest hardware to run Qwen3 8B?
In Q4, the cheapest option in our catalogue to run Qwen3 8B is RTX 3060 12 GB, about €1,180 for the complete machine.
How fast does Qwen3 8B run?
Expect around 39 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 | ~39 tok/s | €1,180 |
| RTX 4060 Ti 16 GB | 1 is enough | ~31 tok/s | €1,350 |
| Radeon RX 7800 XT 16 GB | 1 is enough | ~68 tok/s | €1,400 |
| RTX 4070 12 GB | 1 is enough | ~55 tok/s | €1,450 |
| RTX 3090 24 GB (used) | 1 is enough | ~102 tok/s | €1,650 |
| RTX 4070 Ti SUPER 16 GB | 1 is enough | ~73 tok/s | €1,750 |
| Radeon RX 7900 XTX 24 GB | 1 is enough | ~104 tok/s | €1,850 |
| RTX 4080 SUPER 16 GB | 1 is enough | ~80 tok/s | €1,950 |
| RTX 5080 16 GB | 1 is enough | ~104 tok/s | €2,100 |
| Mini-PC Ryzen AI Max+ 395 128 GB | 1 is enough | ~28 tok/s | €2,300 |
| RTX 4090 24 GB | 1 is enough | ~109 tok/s | €2,900 |
| RTX 5090 32 GB | 1 is enough | ~194 tok/s | €3,500 |
| NVIDIA DGX Spark 128 GB | 1 is enough | ~30 tok/s | €4,000 |
| Mac Studio M4 Max 128 GB | 1 is enough | ~59 tok/s | €4,400 |
| RTX 6000 Ada 48 GB | 1 is enough | ~104 tok/s | €7,900 |
| RTX PRO 6000 Blackwell 96 GB | 1 is enough | ~194 tok/s | €9,900 |
| Mac Studio M3 Ultra 512 GB | 1 is enough | ~89 tok/s | €11,500 |
| L40S 48 GB | 1 is enough | ~94 tok/s | €13,500 |
| A100 80 GB | 1 is enough | ~221 tok/s | €17,000 |
| H100 80 GB | 1 is enough | ~363 tok/s | €34,000 |
| H200 141 GB | 1 is enough | ~521 tok/s | €39,000 |
| B200 192 GB | 1 is enough | ~868 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 3060 12 GB · €1,180