Qwen3 4B hardware requirements
4.02B parameters. Needs about 4.2 GB at Q4_K_M with 8K context. Model card.
Memory needed at each quantization
| Quant | Quality | 8K context | 32K context |
|---|---|---|---|
| FP16 | Full quality | 9.8 GB | 13.5 GB |
| Q8_0 | Near-lossless | 6.1 GB | 9.7 GB |
| Q6_K | Excellent | 5.1 GB | 8.7 GB |
| Q5_K_M | Very good | 4.7 GB | 8.3 GB |
| Q4_K_M | Good, most popular | 4.2 GB | 7.9 GB |
| Q3_K_M | Noticeably degraded | 3.8 GB | 7.4 GB |
| Q2_K | Heavily degraded | 3.3 GB | 6.9 GB |
Run it with Ollama: ollama run qwen3:4b (downloads Q4_K_M by default).
Run it in the cloud
Don't want to buy hardware? Run Qwen3 4B on a rented GPU today.
Rent a cloud GPU by the hour. No upfront cost, cancel any time.
Every GPU and Mac checked
| GPU | Runs? | Best fit |
|---|---|---|
| NVIDIA RTX 3060 12GB | Yes | FP16 · ~9.8 GB · ~27 tok/s |
| NVIDIA RTX 3070 | Yes | Q8_0 · ~6.1 GB · ~63 tok/s |
| NVIDIA RTX 3080 10GB | Yes | FP16 · ~9.8 GB · ~57 tok/s |
| NVIDIA RTX 3090 | Yes | FP16 · ~9.8 GB · ~70 tok/s |
| NVIDIA RTX 4060 | Yes | Q8_0 · ~6.1 GB · ~38 tok/s |
| NVIDIA RTX 4060 Ti 16GB | Yes | FP16 · ~9.8 GB · ~21 tok/s |
| NVIDIA RTX 5060 Ti 16GB | Yes | FP16 · ~9.8 GB · ~33 tok/s |
| NVIDIA RTX 4070 | Yes | FP16 · ~9.8 GB · ~38 tok/s |
| NVIDIA RTX 4070 Ti Super | Yes | FP16 · ~9.8 GB · ~50 tok/s |
| NVIDIA RTX 4080 | Yes | FP16 · ~9.8 GB · ~54 tok/s |
| NVIDIA RTX 4090 | Yes | FP16 · ~9.8 GB · ~75 tok/s |
| NVIDIA RTX 5070 | Yes | FP16 · ~9.8 GB · ~50 tok/s |
| NVIDIA RTX 5070 Ti | Yes | FP16 · ~9.8 GB · ~67 tok/s |
| NVIDIA RTX 5080 | Yes | FP16 · ~9.8 GB · ~72 tok/s |
| NVIDIA RTX 5090 | Yes | FP16 · ~9.8 GB · ~134 tok/s |
| AMD Radeon RX 7800 XT | Yes | FP16 · ~9.8 GB · ~47 tok/s |
| AMD Radeon RX 7900 XTX | Yes | FP16 · ~9.8 GB · ~72 tok/s |
| Apple M4 (16GB) | Yes | FP16 · ~9.8 GB · ~9 tok/s |
| Apple M1 (16GB) | Yes | FP16 · ~9.8 GB · ~5 tok/s |
| Apple M2 Pro (32GB) | Yes | FP16 · ~9.8 GB · ~15 tok/s |
| Apple M3 Max (64GB) | Yes | FP16 · ~9.8 GB · ~30 tok/s |
| Apple M4 Pro (48GB) | Yes | FP16 · ~9.8 GB · ~20 tok/s |
| Apple M4 Max (128GB) | Yes | FP16 · ~9.8 GB · ~41 tok/s |
| Apple M2 Ultra (192GB) | Yes | FP16 · ~9.8 GB · ~60 tok/s |