bge-m3

About bge-m3

BGE M3 is BAAI's multilingual embedding model, supporting semantic search across 100+ languages with a single model. It generates embeddings that capture meaning regardless of language, enabling cross-lingual retrieval where queries in one language find relevant content in others. BGE M3 achieves strong performance on multilingual embedding benchmarks, demonstrating effective cross-lingual transfer. The model supports multiple retrieval modes including dense, sparse, and multi-vector representations. Its open weights enable self-hosting for organizations with multilingual content and data privacy requirements. For developers building global search applications, BGE M3 provides unified multilingual embedding without separate models per language. It's essential for international content platforms and multilingual knowledge bases.

Model Specifications

💰 Real-World Cost Examples

Estimated monthly costs for common use cases

Other Model Lineup

Compare all models from Other to find the best fit

Similar Models from Other Providers

Cross-brand alternatives with similar capabilities

Qwen3 Embedding 8B

Alibaba Qwen

Input: $0.01

Output: Free

Context: 32k

Qwen3 Embedding 8B

Alibaba Qwen

Input: $0.01

Output: Free

Context: 32k

Text Embedding 3 Small

OpenAI

Input: $0.02

Output: Free

Context: 8k

Qwen3 Embedding 4B

Alibaba Qwen

Input: $0.02

Output: Free

Context: 33k

💡 Cheaper Alternatives

Same Brand (Other)

🚀 Quick Start

Get started with bge-m3 API

from openai import OpenAI

client = OpenAI(
    base_url="https://api.provider.com/v1",
    api_key="YOUR_API_KEY"
)

response = client.chat.completions.create(
    model="baai/bge-m3-20251117",
    messages=[
        {"role": "user", "content": "Hello!"}
    ]
)
print(response.choices[0].message.content)

Continue Your Decision