by Baai·Released 18 November 2025
BGE-M3 turns text in several languages into a fingerprint of its meaning, so that search finds text that means the same thing. Each piece of text can run to about 8 pages, so whole long documents are handled in one piece. Use it for searching a multilingual library or matching questions to long passages.
You pay for what each request uses. No subscription, no minimum.
One page, the ordinary use
Every rate
| Meter | ↑ you send | ↓ comes back |
|---|---|---|
| Text tokens | 3,33 DA |
per 1M tokens
Two companies run this model. OpenDunes sends each request to one of them, picked by price and speed. If it fails, the next one takes it. DA per 1M tokens: ↑ what you send.
| Provider | ↑ text | Latency | Uptime |
|---|---|---|---|
| Parasail | 3,33 DA | … | … |
| DeepInfraUnited States | 3,33 DA | … | … |
You pay the price of the provider that served your request. Prices read 15:19 · speed over the last 30 minutes · uptime over 3 days.
Swap the base URL and the key. Your OpenAI code runs unchanged.
import os
from openai import OpenAI
client = OpenAI(
base_url="https://opendunes.com/v1",
api_key=os.environ["OPENDUNES_API_KEY"],
)
resp = client.embeddings.create(
model="baai/bge-m3",
input="Delivery to Oran takes two days.",
)
print(resp.data[0].embedding[:5])Runs this request once on your balance.
About 0,0033 DA per page. ↑ 3,33 DA per 1M tokens.
No. You pay per use from your balance, with no subscription. Run it once from Call it in code, at the same price.
BGE-M3 turns text in several languages into a fingerprint of its meaning, so that search finds text that means the same thing. Each piece of text can run to about 8 pages, so whole long documents are handled in one piece. Use it for searching a multilingual library or matching questions to long passages.
Deposit dinars with your Edahabia or CIB card. Every model, bge-m3 included, is paid from that balance. No foreign card, no conversion.
BAAI says it supports more than 100 working languages but does not list them on the card. Arabic, French and Darja are not named.
Up to 8 194 tokens. Split anything longer.
1 024
Baai. Released 18 November 2025.
BGE Base v1.5 turns English text into a fingerprint of its meaning, so that search finds text that means the same thing; each piece of text can be about half a page. Send sentences or paragraphs, store the fingerprints, and compare them to find documents, FAQs or duplicates that mean the same. It is the lighter English choice, with smaller fingerprints to store than BGE Large.
BGE Large v1.5 turns English sentences, paragraphs and short documents into a fingerprint of their meaning, so that search finds text that means the same thing; each piece of text can be about half a page. Its fingerprints are larger than BGE Base's, which gives finer matching in English at the cost of more storage. It does not cover other languages.
From what its maker and two benchmark sites publish, and where it stands on OpenDunes. Read 7 Nov 2026.
Open weights
The model's weights are public.
Search an Arabic and French knowledge base with one index
Match a French question to an English answer
Index long policy documents for retrieval