// BGE embeddings & rerankers · Embedding
bge-m3: VRAM & local deployment
Embedding model published by BAAI.
released 2024-01-27licence: Commercial OKmemory figures are estimates
01Specs
| Publisher | BAAI · BAAI |
|---|---|
| Family / series | BGE embeddings & rerankers · bge-m3 |
| Type | Embedding (sentence-similarity) |
| Parameters | not stated in official metadata |
| Licence | MIT · Commercial OKPermissive licence (Apache-2.0 / MIT / CC-BY / OpenMDW): commercial use allowed. |
| Released | 2024-01-27 (Hugging Face repo created) |
02How much VRAM it needs (estimate)
Embedding models are small; memory grows with batch size and sequence length.
No estimate: the official metadata does not give a parameter count. Check the model card for tested hardware.
Bits per weight: BF16 16, FP8 8.5, 4-bit 4.85. “Fits” = total ≤ 90% of memory; Mac GPU can use ~2/3 of unified memory (≤36GB) or ~3/4 (larger). For MoE, all experts must be in memory, so size by total parameters. Method →
03How to run it
Only runtimes the publisher or the runtime’s official library names for this model.
- Transformers — mentioned in the official model card
- Sentence Transformers — mentioned in the official model card
- Ollama — in the official Ollama library · bge-m3
ollama run bge-m304Official repositories & quantised builds
1 official repo(s) from BAAI.
- bge-m3 main
05Other BGE embeddings & rerankers models
All 8 →| Model | Params | 4-bit total | Released |
|---|---|---|---|
| bge-code-v1 | 1.54B | 1.5 GB | 2025-05-15 |
| BGE-VL-base | 150M | 0.6 GB | 2025-02-25 |
| BGE-VL-large | 428M | 0.8 GB | 2025-02-25 |
| bge-en-icl | 7.11B | 5.2 GB | 2024-07-25 |
| bge-multilingual-gemma2 | 9.24B | 6.7 GB | 2024-07-25 |
| bge-reranker-v2.5-gemma2-lightweight | 9.24B | 6.7 GB | 2024-07-25 |
| bge-reranker-v2-m3 | 568M | 0.9 GB | 2024-03-15 |
06Sources
Data checked 2024-07-03. Parameters, context and licence are copied from the official repository; memory figures are our estimates.