Open-weight local model
Qwen 3 30B A3B
MoE model: all quantized weights must fit, while active parameters affect calculated speed.
- Parameters
- 30.532B
- Recommended quantization
- Q4_K_M
- Weight estimate
- 18.8GB
- Maximum context
- 40K
Quantization memory at the default context
| Quantization | Weights | KV cache | Total VRAM |
|---|---|---|---|
| FP16 | 61.50GB | 0.81GB | 68.46GB |
| Q8_0 | 33.50GB | 0.81GB | 37.66GB |
| Q6_K | 26.05GB | 0.81GB | 29.46GB |
| Q5_K_M | 22.04GB | 0.81GB | 25.05GB |
| Q4_K_M | 18.80GB | 0.81GB | 21.49GB |
| Q3_K_M | 15.60GB | 0.81GB | 17.97GB |
| Q2_K | 10.02GB | 0.81GB | 11.83GB |
GPUs to check for this model
- Can RTX 3090 24GB run it?
- Can RTX 4090 24GB run it?
- Can AMD Radeon RX 7900 XTX 24GB run it?
- Can Apple MacBook Pro M4 Max (32-core GPU) 36GB run it?
- Can Apple MacBook Pro M5 Pro 36GB run it?
- Can Apple MacBook Pro M3 Max (30-core GPU) 36GB run it?
- Can Apple MacBook Pro M3 Pro (14-core GPU) 36GB run it?
- Can Apple MacBook Pro M3 Pro (18-core GPU) 36GB run it?
- Can RTX 5090 32GB run it?
- Can RTX 5090 D 32GB run it?