Skip to content
Get an API key

Live · serverless · 3 regions

BGE-M3

by BAAI

A multilingual embedding model that returns 1024-dimension vectors for dense, sparse and multi-vector search.

  • embeddings
  • BF16
Serverless priceUSD
Input
$0.01per 1M tokens
Output
—returns vectors, not tokens

Billed per token, no minimums.

Reference

Specifications

Model ID
bge-m3
Endpoint
POST /v1/embeddings
Publisher
BAAI
Tasks
embeddings
Served context
8K tokens
Quantization
BF16
p50 response
18 mslast 24 hours
Regions
us-east · us-west · eu-west
Dedicated
On request

Last 24 hours

Latency

Median of each 15-minute step on the shared pool, measured at the API edge, for the whole response: there is no token stream to time.

Response time · ms

18 msmedian

p95 23 ms

Show the data
Median of each 15-minute step over the last 24 hours, newest first
StepResponse time
15m ago19 ms
30m ago18 ms
45m ago18 ms
1h ago19 ms
1h 15m ago18 ms
1h 30m ago18 ms
1h 45m ago18 ms
2h ago19 ms
2h 15m ago19 ms
2h 30m ago25 ms
2h 45m ago18 ms
3h ago18 ms
3h 15m ago18 ms
3h 30m ago18 ms
3h 45m ago19 ms
4h ago18 ms
4h 15m ago19 ms
4h 30m ago18 ms
4h 45m ago18 ms
5h ago19 ms
5h 15m ago18 ms
5h 30m ago18 ms
5h 45m ago18 ms
6h ago19 ms
6h 15m ago18 ms
6h 30m ago23 ms
6h 45m ago19 ms
7h ago18 ms
7h 15m ago19 ms
7h 30m ago18 ms
7h 45m ago18 ms
8h ago19 ms
8h 15m ago18 ms
8h 30m ago19 ms
8h 45m ago18 ms
9h ago18 ms
9h 15m ago17 ms
9h 30m ago18 ms
9h 45m ago18 ms
10h ago17 ms
10h 15m ago17 ms
10h 30m ago18 ms
10h 45m ago17 ms
11h ago18 ms
11h 15m ago17 ms
11h 30m ago17 ms
11h 45m ago18 ms
12h ago25 ms
12h 15m ago18 ms
12h 30m ago17 ms
12h 45m ago17 ms
13h ago17 ms
13h 15m ago16 ms
13h 30m ago18 ms
13h 45m ago18 ms
14h ago16 ms
14h 15m ago23 ms
14h 30m ago21 ms
14h 45m ago17 ms
15h ago16 ms
15h 15m ago17 ms
15h 30m ago17 ms
15h 45m ago17 ms
16h ago18 ms
16h 15m ago22 ms
16h 30m ago17 ms
16h 45m ago18 ms
17h ago18 ms
17h 15m ago17 ms
17h 30m ago18 ms
17h 45m ago18 ms
18h ago17 ms
18h 15m ago17 ms
18h 30m ago18 ms
18h 45m ago17 ms
19h ago17 ms
19h 15m ago18 ms
19h 30m ago18 ms
19h 45m ago17 ms
20h ago18 ms
20h 15m ago25 ms
20h 30m ago17 ms
20h 45m ago17 ms
21h ago18 ms
21h 15m ago19 ms
21h 30m ago18 ms
21h 45m ago18 ms
22h ago18 ms
22h 15m ago18 ms
22h 30m ago25 ms
22h 45m ago18 ms
23h ago18 ms
23h 15m ago18 ms
23h 30m ago18 ms
23h 45m ago19 ms
24h ago19 ms

API

Make a request

The same request shape as every other model on /v1/embeddings. Switching between them is a change to one string.

curl https://api.weir.example/v1/embeddings \
  -H "Authorization: Bearer $WEIR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "bge-m3",
    "input": "Where is my order?"
  }'
POST /v1/embeddingsapi.weir.example

Dedicated

Need reserved capacity?

BGE-M3 runs on the shared serverless pool. Dedicated replicas for it are set up on request, with the same API and the same model ID.

Talk to an engineer

Compare

Similar models