Qwen3-Embedding 0.6B
Alibaba · Embedding · 0.6B · 33k context · Released 5 June 2025
Commercial use permitted
Open weights
Runs on CPU
Apple Silicon
The smallest model in Alibaba's Qwen3-Embedding family, strong on multilingual retrieval for its size while staying light enough to run on a CPU. Apache 2.0, with user-selectable output dimensions.
Strengths
- Strong multilingual retrieval quality for a sub-1B model
- Selectable output dimensions, from 32 up to 1024
- Apache 2.0, and small enough for CPU use
Weaknesses
- The larger 4B and 8B siblings retrieve better if you have the memory
- Newer than nomic-embed and BGE-M3, so less battle-tested in tooling
- As always, a domain-tuned embedder may beat it on narrow content
Hardware requirements
| Quantisation | Approx. VRAM | Notes |
|---|---|---|
| Q8_0 | ~0.7GB | Light enough for CPU, minimal quality loss |
| FP16 | ~1.2GB | Full precision |
Also runs on CPU (slower). Optimised builds available for Apple Silicon.
Licence
Apache 2.0 — read the licence
Availability
Recommended for
- Multilingual RAG on modest hardware
- Setups that benefit from adjustable embedding dimensions
- A current, permissively licensed embedder
Run it with
Catalogue entry last verified 30 July 2026. Specifications change; verify anything you are about to spend money on.