OpenLuxBGE-M3 API
BGE-M3 is a multilingual embedding model built for retrieval. On OpenLux it is metered at $0.0035 per million tokens — the cheapest line item in the catalogue, and the one that often decides how good your RAG answers are.
Pay-as-you-go rate
$0.0035 / 1M tokens
Billed in USD, metered per request with real-time cost in the console.
Access
One OpenLux key and the OpenAI-compatible endpoint: set base_url to https://openlux.lol/v1 and pass bge-m3 as the model.
Get started
Create a key in the console, then follow the five-minute docs for streaming, tools and error codes.
What BGE-M3 delivers
Embeddings are the retrieval half of a RAG pipeline: text is converted to vectors, vectors are compared, and the model above receives the passages that actually matter. BGE-M3 handles that step across languages, which means a question in one language can retrieve documents written in another.
The economics are unusual — at $0.0035 per million input tokens, embedding an entire documentation set costs pennies, and re-embedding the same corpus after every content change stays trivial. That makes it reasonable to reindex often rather than settle for a stale index.
It runs on the same endpoint as the generation models, so ingestion, retrieval and answering can all live in one pipeline with one key, one rate card and one place to watch cost.
Call BGE-M3 in minutes
Every OpenLux model speaks the OpenAI protocol. Reuse your existing SDKs and replace two values: the base URL and the API key.
cURL
curl https://openlux.lol/v1/chat/completions \
-H "Authorization: Bearer $OPENLUX_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "bge-m3",
"messages": [{"role": "user", "content": "OpenLux charges per request and shows the cost of every call in the console."}]
}'Python (OpenAI SDK)
from openai import OpenAI
client = OpenAI(
api_key="sk-openlux-...",
base_url="https://openlux.lol/v1",
)
resp = client.chat.completions.create(
model="bge-m3",
messages=[{"role": "user", "content": "OpenLux charges per request and shows the cost of every call in the console."}],
)
print(resp.choices[0].message.content)Node.js (OpenAI SDK)
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.OPENLUX_API_KEY,
baseURL: "https://openlux.lol/v1",
});
const resp = await client.chat.completions.create({
model: "bge-m3",
messages: [{ role: "user", content: "OpenLux charges per request and shows the cost of every call in the console." }],
});
console.log(resp.choices[0].message.content);Where BGE-M3 fits
Multilingual vectors
One embedding space for multiple languages, so cross-language retrieval works without duplicate indexes.
Long-context chunks
Handles substantial passages per request, which keeps chunking decisions simple during ingestion.
Semantic search
Ranks by meaning rather than keyword overlap, which is what lets a natural-language question find the right document.
Negligible cost
At $0.0035 per million tokens, reindexing a knowledge base is cheap enough to do on every deploy.
RAG pipelines
Embed documentation, tickets or contracts once and feed the relevant passages to any OpenLux text model.
Semantic search
Replace brittle keyword search with meaning-based retrieval in a product or an internal tool.
Deduplication and clustering
Group similar records, spot near-duplicates and route content automatically using vectors.
How bge-m3 is priced on OpenLux
BGE-M3 is metered pay as you go at $0.0035 / 1M tokens. There is no subscription requirement for API access: top up a balance, call the model, and see the cost of every request in the console.
| Model | Provider | Rate | Discount |
|---|---|---|---|
| bge-m3 (this model) | AIGC | $0.0035 / 1M tokens | — |
| jina-reranker-v3 | AIGC | $0.0042 / 1M tokens | — |
| deepseek-v4 | DeepSeek | $0.0613 / 1M tokens | -90% |
| qwen3-max | Qwen | $0.0784 / 1M tokens | -88% |
Rates follow official provider pricing and are shown in USD. See the full model catalogue and pricing plans for credit allowances.
BGE-M3 questions
How much does BGE-M3 cost on OpenLux?
$0.0035 per 1 million input tokens — metered pay as you go, with no minimum spend. Embedding a large corpus typically costs a few cents.
How do I call BGE-M3?
Send "model": "bge-m3" with your text to https://openlux.lol/v1 using your OpenLux API key, then store the returned vectors in your vector store.
Does it work across languages?
Yes — it is a multilingual model, so text in one language can be matched against documents in another within the same index.
Which ranking step should I pair it with?
BGE-M3 retrieves broadly and jina-reranker-v3 reorders the candidates. Both are on the same key, and the pair together costs a fraction of a cent per query.
Start calling BGE-M3 today
Create an OpenLux key and send your first request in minutes. Compare BGE-M3 with the rest of the catalogue on the same endpoint and the same bill.