BGE-M3
BAAI · 567M · up to 8K context · mit
A multilingual embedding model small enough to run on any Mac without noticing it in memory. It supports dense, sparse, and multi-vector retrieval in one model, which covers most local RAG setups without needing a second embedding pass.
On a M4 Pro with 48 GB
macOS9.6 GBweights1.1 GBKV cache0.0 GBheadroom37.3 GB
- Weights
- 1.1 GB
- KV cache
- 0.0 GBestimated shape
- Usable RAM
- 38.4 GB
- Headroom
- 37.3 GB
- Generation
- 188 tok/sest.
- Prompt processing
- 1593 tok/sest.
- First token (1K prompt)
- 643 msest.
Decode speed is bandwidth ÷ active weight bytes, at 273 GB/s and 78% efficiency. These are modelled figures, not measurements.
Every quantization, on your Mac
| Quant | Weights | + KV | Verdict | Speed est. | Repo |
|---|---|---|---|---|---|
| q4 | 0.3 GB | 0.3 GB | FITS · 38.1 GB FREE | 668 tok/s | bge-m3-mlx-4bit |
| q6 | 0.5 GB | 0.5 GB | FITS · 37.9 GB FREE | 462 tok/s | bge-m3-mlx-6bit |
| q8 | 0.6 GB | 0.6 GB | FITS · 37.8 GB FREE | 353 tok/s | bge-m3-mlx-8bit |
| bf16 | 1.1 GB | 1.1 GB | FITS · 37.3 GB FREE | 188 tok/s | bge-m3-mlx-fp16 |
We never host weights. Every link goes to Hugging Face.
Run it
Chat
mlx_lm.chat --model mlx-community/bge-m3-mlx-fp16Serve an OpenAI-compatible endpoint
mlx_lm.server --model mlx-community/bge-m3-mlx-fp16 --port 8080How far can you push the context
2K tokensFITS · 37.3 GB FREE
cache 0.0 GB
8K tokensFITS · 37.3 GB FREE
cache 0.0 GB
Which Macs run this
| Chip | Smallest RAM that fits | Speed est. |
|---|---|---|
| M1 | 8 GB | 43.2 tok/s |
| M1 Pro | 16 GB | 134 tok/s |
| M1 Max | 32 GB | 279 tok/s |
| M1 Ultra | 64 GB | 571 tok/s |
| M2 | 8 GB | 64.4 tok/s |
| M2 Pro | 16 GB | 136 tok/s |
| M2 Max | 32 GB | 282 tok/s |
| M2 Ultra | 64 GB | 578 tok/s |
| M3 | 8 GB | 64.4 tok/s |
| M3 Pro | 18 GB | 99.2 tok/s |
| M3 Max (14-core CPU) | 36 GB | 206 tok/s |
| M3 Max (16-core CPU) | 48 GB | 282 tok/s |
| M3 Ultra | 96 GB | 599 tok/s |
| M4 | 16 GB | 78.3 tok/s |
| M4 Pro | 24 GB | 188 tok/s |
| M4 Max (14-core CPU) | 36 GB | 293 tok/s |
| M4 Max (16-core CPU) | 48 GB | 395 tok/s |
| M5 | 16 GB | 101 tok/s |
| M5 Prounverified | 24 GB | 214 tok/s |
| M5 Maxunverified | 36 GB | 449 tok/s |