MXFP4 Open on Hugging Face
Parameters
7.3B
Weights (est.)
≈ 4.1 GB
Mac memory (est.)
8 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.
Run it
With mlx-lm on an Apple silicon Mac:
pip install mlx-lmmlx_lm.generate --model mlx-community/Olmo-3-7B-Instruct-mxfp4-QAT --prompt "Hello"
Details
- Base model
- allenai/Olmo-3-7B-Instruct (quantized)
- Vendor
- Ai2 · OLMo
- Task
- Text generation
- Quantisation
- MXFP4
- Licence
- —
- Downloads
- 458 in the last 30 days · 1.8K all time · 4 likes
- Published
- 2026-04-22 · updated 2026-04-23 (6mo ago)
- Tags
- olmo3, mlx-lm-lora
Other MLX builds of allenai/Olmo-3-7B-Instruct
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Olmo-3-7B-Instruct-4bit | 4-bit | ≈ 4.1 GB | 8 GB+ | 768 |
| Olmo-3-7B-Instruct-mxfp4-QAT (this one) | MXFP4 | ≈ 4.1 GB | 8 GB+ | 458 |
| Olmo-3-7B-Instruct-6bit | 6-bit | ≈ 5.9 GB | 16 GB+ | 17 |
| Olmo-3-7B-Instruct-8bit | 8-bit | ≈ 7.8 GB | 16 GB+ | 415 |
| Olmo-3-7B-Instruct-bf16 | bf16 | ≈ 14.6 GB | 24 GB+ | 33 |