4-bit Open on Hugging Face
Parameters
—
Weights (est.)
—
Mac memory (est.)
—
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/Qwen3-8B (quantized) · Qwen3 8B on the radar →
- Vendor
- Qwen
- Task
- Embeddings
- Quantisation
- 4-bit
- Licence
- apache-2.0
- Downloads
- 104 in the last 30 days · 373 all time · 1 likes
- Published
- 2026-09-29 · updated 2026-09-29 (1mo ago)
- Tags
- qwen3, apple-silicon, contrastive, embeddings, verifier
Other MLX builds of Qwen/Qwen3-8B
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Qwen3-8B-DFly-MLX | — | ≈ 2.3 GB | 8 GB+ | 13 |
| Qwen3-8B-3bit | 3-bit | ≈ 3.6 GB | 8 GB+ | 191 |
| Qwen3-8B-4bit | 4-bit | ≈ 4.6 GB | 8 GB+ | 33K |
| Qwen3-8B-4bit-DWQ | 4-bit DWQ | ≈ 4.6 GB | 8 GB+ | 351 |
| Qwen3-8B-4bit-DWQ-053125 | 4-bit DWQ | ≈ 4.6 GB | 8 GB+ | 214 |
| Qwen3-8B-4bit-AWQ | 4-bit | ≈ 4.6 GB | 8 GB+ | 193 |
| Molmo2-8B-4bit | 4-bit | ≈ 6.4 GB | 16 GB+ | 201 |
| Qwen3-8B-6bit | 6-bit | ≈ 6.7 GB | 16 GB+ | 345 |
| MolmoPoint-8B-4bit | 4-bit | ≈ 7.2 GB | 16 GB+ | 108 |
| MolmoPoint-8B-nvfp4 | 4-bit | ≈ 7.2 GB | 16 GB+ | 50 |
| MolmoPoint-8B-mxfp4 | MXFP4 | ≈ 7.2 GB | 16 GB+ | 38 |
| Molmo2-8B-5bit | 5-bit | ≈ 7.4 GB | 16 GB+ | 41 |
| MolmoPoint-8B-5bit | 5-bit | ≈ 8.1 GB | 16 GB+ | 45 |
| Molmo2-8B-6bit | 6-bit | ≈ 8.3 GB | 16 GB+ | 46 |
| Qwen3-8B-8bit | 8-bit | ≈ 8.7 GB | 16 GB+ | 953 |
| MolmoPoint-8B-6bit | 6-bit | ≈ 9.0 GB | 16 GB+ | 52 |
| Molmo2-8B-8bit | 8-bit | ≈ 10.2 GB | 16 GB+ | 86 |
| MolmoPoint-8B-8bit | 8-bit | ≈ 10.7 GB | 16 GB+ | 55 |
| MolmoPoint-8B-mxfp8 | MXFP8 | ≈ 10.7 GB | 16 GB+ | 34 |
| Qwen3-8B-bf16 | bf16 | ≈ 16.4 GB | 24 GB+ | 559 |
| Molmo2-8B-fp16 | fp16 | ≈ 17.3 GB | 24 GB+ | 41 |
| MolmoPoint-8B-fp16 | fp16 | ≈ 17.4 GB | 24 GB+ | 34 |
| CLM-v0.1-8B-MLX-8bit | 8-bit | — | — | 635 |
| CLM-v0.1-8B-MLX-4bit (this one) | 4-bit | — | — | 104 |
| CLM-v0.1-8B-MLX-5bit | 5-bit | — | — | 46 |