Granite Embedding Small English R2
Aug 2025
Granite Embedding Small English R2Compare inon
Granite Embedding Small English R2 measured on 25 of 194 Embeddings benchmarks25
BenchmarkByScoreRank
No benchmark matches
self-host onlyprice
53.9MTEB English v2
Aug 2025released
384Dimensions
8K tokensMax input
enLanguages
Rankingconfiguration unstated?
Full ranking →···
70Mdbr Leaf Ir41.271Granite Embedding Small English R240.172Mdbr Leaf MT39.3Strong at
Alternatives?
ModelScoreCostPriceWhat changes
BetterBge Base En v1.5unstatedOpen+9%—$0.005Max input: 8K tokens → 1K tokens+1
OpenNemotron 3 Embed 8B BF16+77%——Memory (4-bit): 0 GB → 4.8 GB
BetterQwen 3 Embedding 8BunstatedOpen+67%—$0.01Languages: not verified+1BetterBge M3unstatedOpen+37%—$0.01Languages: not verified+1iBetterMultilingual E5 LargeunstatedOpen+35%—$0.01Max input: 8K tokens → 1K tokens+1ModelScoreCostPriceWhat changes
Nothing here yet.
ModelScoreCostPriceWhat changes
Bge Base En v1.5unstatedOpen+9%—$0.005Max input: 8K tokens → 1K tokens+1Qwen 3 Embedding 8BunstatedOpen+67%—$0.01Languages: not verified+1Bge M3unstatedOpen+37%—$0.01Languages: not verified+1iMultilingual E5 LargeunstatedOpen+35%—$0.01Max input: 8K tokens → 1K tokens+1iMultilingual E5 Large InstructunstatedOpen+22%—$0.01Max input: 8K tokens → 1K tokens+1ModelScoreCostPriceWhat changes
Nemotron 3 Embed 8B BF16Open+77%——Memory (4-bit): 0 GB → 4.8 GBOOcten Embedding 4BOpen+70%——Memory (4-bit): 0 GB → 2.4 GBOOcten Embedding 4B INT8Open+69%——Memory (4-bit): 0 GB → 2.4 GBOOcten Embedding 8BOpen+69%——Memory (4-bit): 0 GB → 4.5 GBNV Embed v2unstatedOpen+68%——Memory (4-bit): 0 GB → 4.7 GBModelScoreCostPriceWhat changes
Granite Embedding 311M Multilingual R2unstatedOpen+17%——Memory (4-bit): 0 GB → 0.2 GBGranite Embedding English R2unstatedOpen+10%——Memory (4-bit): 0 GB → 0.1 GBGranite Embedding 97M Multilingual R2unstatedOpen+3%——Memory (4-bit): 0 GB → 0.1 GBGranite Embedding 107M MultilingualunstatedOpenn/a——Max input: 8K tokens → 1K tokens+1Granite Embedding 125M EnglishunstatedOpenn/a——Max input: 8K tokens → 1K tokens+1Granite Embedding 278M MultilingualunstatedOpenn/a——Max input: 8K tokens → 1K tokens+1Granite Embedding 30M EnglishunstatedOpenn/a——Max input: 8K tokens → 1K tokensAll IBM models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceApache-2.0free for commercial use
Benchmarks
in Embeddings & RetrievalBEIR · Retrieval · English50.9#97
FAQ
How much does Granite Embedding Small English R2 cost?›
There is no per-unit API price: self-host only.
Is Granite Embedding Small English R2 free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Granite Embedding Small English R2, and what is it best at?›
#69 of 200 in Embeddings.
Is Granite Embedding Small English R2 better than Mdbr Leaf Ir?›
On RTEB Health beta · Mean over tasks · multilingual (4 languages), Granite Embedding Small English R2 scores 40.1 vs 41.2 — lower.
When was Granite Embedding Small English R2 released? Is there a newer version?›
Released 2025-08-15. It is the current version in its line.
Is Granite Embedding Small English R2 open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Granite Embedding Small English R2 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
What is Granite Embedding Small English R2?›
Granite Embedding Small English R2 is an open-weights embeddings & retrieval model from IBM, released Aug 2025.
What are the best alternatives to Granite Embedding Small English R2?›
Bge Base En v1.5 (better), Nemotron 3 Embed 8B BF16 (open weights).