baai
BAAI: bge-m3
baai/bge-m3
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
baai/bge-m3 is a multilingual text embedding model from BAAI that produces 1024-dimensional dense vectors for inputs up to 8192 tokens per its model card; the OpenRouter catalog lists 8194. According to the vendor documentation, the same encoder also supports sparse lexical-weight and multi-vector (ColBERT) retrieval, with coverage of more than 100 languages. Published benchmark results come mainly from the vendor and a community evaluation it cites, so validation on your own workload is advisable. The model card states an MIT license.[1][2]
Strengths
- One encoder supports dense, sparse (lexical-weight), and multi-vector (ColBERT) retrieval modes, enabling hybrid retrieval without a separate lexical model.[2]
- The vendor reports coverage of more than 100 languages and inputs ranging from short sentences to long documents of up to 8192 tokens.[1][2]
- Produces 1024-dimensional dense embeddings; the OpenRouter catalog lists its modality as text to embeddings.[1][2]
Limitations
- It is an embedding model only: it outputs vectors and does not generate text.[1][2]
- Published results are vendor-reported or community evaluations the vendor cites; the vendor itself notes BM25 remains a competitive baseline, especially on long documents, so independent validation on your data is needed.[2]
- The vendor notes that half-precision (fp16) inference speeds computation at the cost of a slight performance degradation.[2]
Where to use it
- A reasonable candidate for multilingual semantic search and retrieval over long documents.[1][2]
- Worth evaluating for hybrid retrieval in RAG pipelines; the vendor suggests pairing retrieval with a re-ranker for higher accuracy.[2]
- Potentially useful where BM25-style lexical weights and dense embeddings are wanted from a single model.[2]
Sources
- [1] OpenRouter catalog · BAAI: bge-m3Sources checked: Nov 13, 2026
- [2] Model documentation · BAAI: bge-m3Sources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
The bge-m3 embedding model encodes sentences, paragraphs, and long documents into a 1024-dimensional dense vector space, delivering high-quality semantic embeddings optimized for multilingual retrieval, semantic search, and large-context applications.
Specifications
- Model ID
- baai/bge-m3
- Canonical ID
- baai/bge-m3-20251117
- Creator
- baai
- Added to OpenRouter
- 18.11.2025
- Context length
- 8,194
- Modality
- text->embeddings
- Input modalities
- text
- Output modalities
- embeddings
- Token pricing
- USD per 1M tokens: in $0.010 / out $0
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | — |
| Coding | No measurement | — |
| Agents | No measurement | — |
| Writing | No measurement | — |
| Reasoning | No measurement | — |
| Document analysis | No measurement | — |
| Vision | No measurement | — |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | output: embeddings |
Evaluations
No evaluations are supplied for this model.
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| Parasail | 8,194 | $0.010 | $0 | — | — | 100% |
| DeepInfrafp32 | 8,192 | $0.010 | $0 | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.