Qwen3 235B-A22B: how much VRAM?
In Q4_K_M with 8k tokens of context, Qwen3 235B-A22B needs about 150 GB of memory. The cheapest option: Mac Studio M3 Ultra 512 GB (€11,500 for the complete machine).
Model sheet
ollama run qwen3:235bWeights on Hugging Face: Qwen/Qwen3-235B-A22B
VRAM by precision and context
| Precision | 4k | 8k | 32k | 128k |
|---|---|---|---|---|
| FP16 | 495 GB | 495 GB | 500 GB | 517 GB |
| FP8 | 248 GB | 249 GB | 253 GB | 270 GB |
| Q8_0 | 263 GB | 263 GB | 268 GB | 285 GB |
| Q6_K | 203 GB | 204 GB | 208 GB | 226 GB |
| Q4_K_M | 149 GB | 150 GB | 154 GB | 171 GB |
| Q3_K_M | 122 GB | 123 GB | 127 GB | 144 GB |
One user. Each simultaneous user adds their own context cache.
Frequently asked questions
How much VRAM does Qwen3 235B-A22B need?
With an 8k-token context and one user, Qwen3 235B-A22B needs about 150 GB in Q4, 263 GB in Q8 and 495 GB in FP16.
What is the cheapest hardware to run Qwen3 235B-A22B?
In Q4, the cheapest option in our catalogue to run Qwen3 235B-A22B is Mac Studio M3 Ultra 512 GB, about €11,500 for the complete machine.
How fast does Qwen3 235B-A22B run?
Expect around 34 tokens per second for one user on Mac Studio M3 Ultra 512 GB in Q4 (estimate based on memory bandwidth).
On which hardware?
| Hardware | Cards | Speed | Machine |
|---|---|---|---|
| Mac Studio M3 Ultra 512 GB | 1 is enough | ~34 tok/s | €11,500 |
| RTX PRO 6000 Blackwell 96 GB | 2 cards | ~111 tok/s | €18,900 |
| A100 80 GB | 2 cards | ~126 tok/s | €28,000 |
| RTX 6000 Ada 48 GB | 4 cards | ~119 tok/s | €30,800 |
| L40S 48 GB | 4 cards | ~107 tok/s | €36,000 |
| B200 192 GB | 1 is enough | ~330 tok/s | €48,000 |
| H100 80 GB | 2 cards | ~207 tok/s | €62,000 |
| H200 141 GB | 2 cards | ~297 tok/s | €72,000 |
| RTX 3060 12 GB | doesn’t fit | — | — |
| RTX 4060 Ti 16 GB | doesn’t fit | — | — |
| Radeon RX 7800 XT 16 GB | doesn’t fit | — | — |
| RTX 4070 12 GB | doesn’t fit | — | — |
| RTX 3090 24 GB (used) | doesn’t fit | — | — |
| RTX 4070 Ti SUPER 16 GB | doesn’t fit | — | — |
| Radeon RX 7900 XTX 24 GB | doesn’t fit | — | — |
| RTX 4080 SUPER 16 GB | doesn’t fit | — | — |
| RTX 5080 16 GB | doesn’t fit | — | — |
| RTX 4090 24 GB | doesn’t fit | — | — |
| Mini-PC Ryzen AI Max+ 395 128 GB | doesn’t fit | — | — |
| RTX 5090 32 GB | doesn’t fit | — | — |
| NVIDIA DGX Spark 128 GB | doesn’t fit | — | — |
| Mac Studio M4 Max 128 GB | doesn’t fit | — | — |
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: 2 × RTX 3060 12 GB · €1,460