Qwen3-32B-3bit
mlx-community/Qwen3-32B-3bit
Uncategorised · mlx
3-bit Open on Hugging Face
Parameters
33B
Weights (est.)
≈ 14.3 GB
Mac memory (est.)
24 GB+
Size is estimated from the parameter count and quantisation. Memory allows a fifth extra for context plus 2 GB for the system; long contexts need more.
Details
- Base model
- Qwen/Qwen3-32B (quantized)
- Vendor
- Qwen
- Task
- Uncategorised
- Quantisation
- 3-bit
- Licence
- apache-2.0
- Downloads
- 45 in the last 30 days · 710 all time · 2 likes
- Published
- 2025-07-01 · updated 2025-07-01 (1y ago)
- Tags
- qwen3, 3bit, quantized
Other MLX builds of Qwen/Qwen3-32B
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Qwen3-32B-3bit (this one) | 3-bit | ≈ 14.3 GB | 24 GB+ | 45 |
| Qwen3-32B-4bit | 4-bit | ≈ 18.4 GB | 24 GB+ | 2.0K |
| Qwen3-30B-4bit-DWQ-053125 | 4-bit DWQ | ≈ 18.4 GB | 24 GB+ | 30 |
| Qwen3-32B-6bit | 6-bit | ≈ 26.6 GB | 32 GB+ | 86 |
| Qwen3-32B-8bit | 8-bit | ≈ 34.8 GB | 48 GB+ | 422 |
| Qwen3-32B-bf16 | bf16 | ≈ 65.5 GB | 96 GB+ | 51 |