BGE-M3 Embeddings

Multilingual embedding model for search and RAG, 8K-token inputs.

Overview

BGE-M3 produces dense embeddings for 100+ languages and long inputs. Call Ollama's /api/embed endpoint.

Key features

• ✓
  100+ languages
• ✓
  8K input length
• ✓
  MIT licence

Use cases

• →
  Semantic search
• →
  RAG pipelines
• →
  Clustering and deduplication

On this template

Runs on the Ollama template, exposing its Ollama API at a private HTTPS address. You also get root SSH and JupyterLab.

OLLAMA_PULL

bge-m3

Quick start

1. Click Deploy and pick an available GPU — only offers with enough VRAM are shown.
2. Wait for the status to turn running. The first start downloads the model, which can take a few minutes.
3. Use the connection details on your rental page:

curl -u edge:<app-token> https://r12-abcd1234-app.gpu.clearlayer.ca/api/generate \
  -d '{"model": "bge-m3", "prompt": "Hello!", "stream": false}'

import requests

r = requests.post("https://r12-abcd1234-app.gpu.clearlayer.ca/api/chat", auth=("edge", "<app-token>"), json={
    "model": "bge-m3", "stream": False,
    "messages": [{"role": "user", "content": "Hello!"}],
})
print(r.json()["message"]["content"])