All tasks / Models
granite-embedding-311m-multilingual-r2
IBM's small Apache-2.0 open-weights embedding models for English and multilingual retrieval.
- IBM
- Open weights
- Released Apr 2026
Where it ranks
- CEmbeddings & retrievalModels · Local · 56% on MTEB(Multilingual, v2)
Facts
- Vendor
- IBM
- Family
- Granite Embedding
- Released
- Apr 2026
- Weights
- Open: download and run it yourself
- License
- Apache-2.0
- Parameters
- 0.31B
- GPU memory to run locally
- ~1 GB (Board lists ~0.6 GB memory use (0.31B parameters).)
- Access
- Free (open weights)
- Last verified
- 19 Oct 2026
Benchmark results
| Benchmark | Result | Entry on the leaderboard | As of | Source |
|---|---|---|---|---|
| MTEB(Multilingual, v2) | 56% | ibm-granite/granite-embedding-311m-multilingual-r2Mean (Task) over all 131 tasks; board rank 73 (Borda); board zero-shot 96%; read from the leaderboard's public API. | 19 Oct 2026 | mteb-leaderboard-backend.hf.space |
Where you can use it
- SOllamaVector databases & engines · Local
Earlier versions
- granite-embedding-english-r2 · Jul 2025