Granite Embedding 278M Multilingual
IBM · ibm-granite/granite-embedding-278m-multilingual
Weights license: Apache 2.0 · code and components may use other terms
Quality
MTEB (Multilingual, v2)
Mean(task) · 0–100 · higher is better · this model outlined
- Color = publisher
Source: MTEB results dataset (independent runs, CC0) · publisher-reported where the model card is the only source · other benchmarks and versions are not mixed in
| MTEB (Multilingual, v2) | mean(task) · 768d | 53.74 | independent run |
|---|---|---|---|
| MTEB (eng, v2) | mean(task) · 768d | 60.36 | independent run |
Reported by the publisher
- MTEB AmazonCounterfactualClassification (en-ext) (en-ext; test; main_score)
- 73.433Classification
- MTEB AmazonCounterfactualClassification (en) (en; test; main_score)
- 71.806Classification
- MTEB AmazonPolarityClassification (default) (test; main_score)
- 67.591Classification
- MTEB AmazonReviewsClassification (en) (en; test; main_score)
- 37.278Classification
- MTEB AppsRetrieval (default) (test; main_score)
- 6.214Retrieval
- MTEB ArguAna (default) (test; main_score)
- 55.244Retrieval
- MTEB ArxivClusteringP2P (default) (test; main_score)
- 43.132Clustering
- MTEB ArxivClusteringS2S (default) (test; main_score)
- 32.934Clustering
Every reported result
- MTEB AskUbuntuDupQuestions (default) (test; main_score)
- 62.344Reranking
- MTEB BIOSSES (default) (test; main_score)
- 81.636STS
- MTEB Banking77Classification (default) (test; main_score)
- 78.062Classification
- MTEB BiorxivClusteringP2P (default) (test; main_score)
- 35.827Clustering
- MTEB BiorxivClusteringS2S (default) (test; main_score)
- 30.391Clustering
- MTEB COIRCodeSearchNetRetrieval (python) (python; test; main_score)
- 88.789Retrieval
- MTEB COIRCodeSearchNetRetrieval (javascript) (javascript; test; main_score)
- 38.778Retrieval
- MTEB COIRCodeSearchNetRetrieval (go) (go; test; main_score)
- 56.296Retrieval
- MTEB COIRCodeSearchNetRetrieval (ruby) (ruby; test; main_score)
- 42.536Retrieval
- MTEB COIRCodeSearchNetRetrieval (java) (java; test; main_score)
- 48.54Retrieval
- MTEB COIRCodeSearchNetRetrieval (php) (php; test; main_score)
- 42.511Retrieval
- MTEB CQADupstackAndroidRetrieval (default) (test; main_score)
- 53.048Retrieval
- MTEB CQADupstackEnglishRetrieval (default) (test; main_score)
- 43.868Retrieval
- MTEB CQADupstackGamingRetrieval (default) (test; main_score)
- 56.369Retrieval
- MTEB CQADupstackGisRetrieval (default) (test; main_score)
- 43.503Retrieval
- MTEB CQADupstackMathematicaRetrieval (default) (test; main_score)
- 32.024Retrieval
- MTEB CQADupstackPhysicsRetrieval (default) (test; main_score)
- 47.316Retrieval
- MTEB CQADupstackProgrammersRetrieval (default) (test; main_score)
- 43.581Retrieval
- MTEB CQADupstackRetrieval (default) (test; main_score)
- 42.743Retrieval
- MTEB CQADupstackRetrieval (default) (test; main_score)
- 42.743Retrieval
- MTEB CQADupstackStatsRetrieval (default) (test; main_score)
- 36.796Retrieval
- MTEB CQADupstackTexRetrieval (default) (test; main_score)
- 31.775Retrieval
- MTEB CQADupstackUnixRetrieval (default) (test; main_score)
- 45.275Retrieval
- MTEB CQADupstackWebmastersRetrieval (default) (test; main_score)
- 42.957Retrieval
- MTEB CQADupstackWordpressRetrieval (default) (test; main_score)
- 36.409Retrieval
- MTEB ClimateFEVER (default) (test; main_score)
- 29.494Retrieval
- MTEB CodeFeedbackMT (default) (test; main_score)
- 31.392Retrieval
- MTEB CodeFeedbackST (default) (test; main_score)
- 67.722Retrieval
- MTEB CodeSearchNetCCRetrieval (python) (python; test; main_score)
- 45.532Retrieval
- MTEB CodeSearchNetCCRetrieval (javascript) (javascript; test; main_score)
- 46.024Retrieval
- MTEB CodeSearchNetCCRetrieval (go) (go; test; main_score)
- 37.555Retrieval
- MTEB CodeSearchNetCCRetrieval (ruby) (ruby; test; main_score)
- 47.549Retrieval
- MTEB CodeSearchNetCCRetrieval (java) (java; test; main_score)
- 45.539Retrieval
- MTEB CodeSearchNetCCRetrieval (php) (php; test; main_score)
- 35.942Retrieval
- MTEB CodeSearchNetRetrieval (python) (python; test; main_score)
- 83.12Retrieval
- MTEB CodeSearchNetRetrieval (javascript) (javascript; test; main_score)
- 70.34Retrieval
- MTEB CodeSearchNetRetrieval (go) (go; test; main_score)
- 86.139Retrieval
- MTEB CodeSearchNetRetrieval (ruby) (ruby; test; main_score)
- 74.736Retrieval
- MTEB CodeSearchNetRetrieval (java) (java; test; main_score)
- 70.89Retrieval
- MTEB CodeSearchNetRetrieval (php) (php; test; main_score)
- 72.930Retrieval
- MTEB CodeTransOceanContest (default) (test; main_score)
- 60.831Retrieval
- MTEB CodeTransOceanDL (default) (test; main_score)
- 32.138Retrieval
- MTEB CosQA (default) (test; main_score)
- 33.452Retrieval
- MTEB DBPedia (default) (test; main_score)
- 34.565Retrieval
- MTEB EmotionClassification (default) (test; main_score)
- 36.450Classification
- MTEB FEVER (default) (test; main_score)
- 84.932Retrieval
- MTEB FiQA2018 (default) (test; main_score)
- 35.663Retrieval
- MTEB HotpotQA (default) (test; main_score)
- 61.513Retrieval
- MTEB ImdbClassification (default) (test; main_score)
- 63.023Classification
- MTEB MIRACLRetrieval (ar) (ar; dev; main_score)
- 64.238Retrieval
- MTEB MIRACLRetrieval (bn) (bn; dev; main_score)
- 68.055Retrieval
- MTEB MIRACLRetrieval (de) (de; dev; main_score)
- 48.123Retrieval
- MTEB MIRACLRetrieval (en) (en; dev; main_score)
- 49.372Retrieval
- MTEB MIRACLRetrieval (es) (es; dev; main_score)
- 49.688Retrieval
- MTEB MIRACLRetrieval (fa) (fa; dev; main_score)
- 50.226Retrieval
- MTEB MIRACLRetrieval (fi) (fi; dev; main_score)
- 67.46Retrieval
- MTEB MIRACLRetrieval (fr) (fr; dev; main_score)
- 49.891Retrieval
- MTEB MIRACLRetrieval (hi) (hi; dev; main_score)
- 46.141Retrieval
- MTEB MIRACLRetrieval (id) (id; dev; main_score)
- 47.229Retrieval
- MTEB MIRACLRetrieval (ja) (ja; dev; main_score)
- 62.81Retrieval
- MTEB MIRACLRetrieval (ko) (ko; dev; main_score)
- 59.216Retrieval
- MTEB MIRACLRetrieval (ru) (ru; dev; main_score)
- 52.349Retrieval
- MTEB MIRACLRetrieval (sw) (sw; dev; main_score)
- 61.271Retrieval
- MTEB MIRACLRetrieval (te) (te; dev; main_score)
- 79.149Retrieval
- MTEB MIRACLRetrieval (th) (th; dev; main_score)
- 73.324Retrieval
- MTEB MIRACLRetrieval (yo) (yo; dev; main_score)
- 68.705Retrieval
- MTEB MIRACLRetrieval (zh) (zh; dev; main_score)
- 52.553Retrieval
- MTEB MSMARCO (default) (dev; main_score)
- 29.867Retrieval
- MTEB MTOPDomainClassification (en) (en; test; main_score)
- 89.900Classification
- MTEB MTOPIntentClassification (en) (en; test; main_score)
- 60.269Classification
- MTEB MassiveIntentClassification (en) (en; test; main_score)
- 63.951Classification
- MTEB MassiveScenarioClassification (en) (en; test; main_score)
- 70.928Classification
- MTEB MedrxivClusteringP2P (default) (test; main_score)
- 31.522Clustering
- MTEB MedrxivClusteringS2S (default) (test; main_score)
- 28.573Clustering
- MTEB MindSmallReranking (default) (test; main_score)
- 30.538Reranking
- MTEB NFCorpus (default) (test; main_score)
- 28.903Retrieval
- MTEB NQ (default) (test; main_score)
- 53.410Retrieval
- MTEB QuoraRetrieval (default) (test; main_score)
- 86.871Retrieval
- MTEB RedditClustering (default) (test; main_score)
- 45.857Clustering
- MTEB RedditClusteringP2P (default) (test; main_score)
- 54.990Clustering
- MTEB SCIDOCS (default) (test; main_score)
- 17.721Retrieval
- MTEB SICK-R (default) (test; main_score)
- 68.745STS
- MTEB STS12 (default) (test; main_score)
- 75.191STS
- MTEB STS13 (default) (test; main_score)
- 77.305STS
- MTEB STS14 (default) (test; main_score)
- 74.948STS
- MTEB STS15 (default) (test; main_score)
- 86.017STS
- MTEB STS16 (default) (test; main_score)
- 81.957STS
- MTEB STS17 (en-tr) (en-tr; test; main_score)
- 44.643STS
- MTEB STS17 (it-en) (it-en; test; main_score)
- 68.276STS
- MTEB STS17 (en-en) (en-en; test; main_score)
- 78.593STS
- MTEB STS17 (en-ar) (en-ar; test; main_score)
- 59.839STS
- MTEB STS17 (fr-en) (fr-en; test; main_score)
- 74.250STS
- MTEB STS17 (nl-en) (nl-en; test; main_score)
- 68.042STS
- MTEB STS17 (es-en) (es-en; test; main_score)
- 69.573STS
- MTEB STS17 (en-de) (en-de; test; main_score)
- 69.718STS
- MTEB STS22 (de-en) (de-en; test; main_score)
- 60.044STS
- MTEB STS22 (en) (en; test; main_score)
- 68.151STS
- MTEB STS22 (es-en) (es-en; test; main_score)
- 80.470STS
- MTEB STS22 (zh-en) (zh-en; test; main_score)
- 72.457STS
- MTEB STS22 (pl-en) (pl-en; test; main_score)
- 70.675STS
- MTEB STSBenchmark (default) (test; main_score)
- 78.408STS
- MTEB SciDocsRR (default) (test; main_score)
- 79.610Reranking
- MTEB SciFact (default) (test; main_score)
- 65.847Retrieval
- MTEB SprintDuplicateQuestions (default) (test; main_score)
- 93.508PairClassification
- MTEB StackExchangeClustering (default) (test; main_score)
- 55.931Clustering
- MTEB StackExchangeClusteringP2P (default) (test; main_score)
- 32.930Clustering
- MTEB StackOverflowDupQuestions (default) (test; main_score)
- 51.776Reranking
- MTEB StackOverflowQA (default) (test; main_score)
- 78.45Retrieval
- MTEB SummEval (default) (test; main_score)
- 32.026Summarization
- MTEB SyntheticText2SQL (default) (test; main_score)
- 47.344Retrieval
- MTEB TRECCOVID (default) (test; main_score)
- 68.573Retrieval
- MTEB Touche2020 (default) (test; main_score)
- 27.306Retrieval
- MTEB ToxicConversationsClassification (default) (test; main_score)
- 62.481Classification
- MTEB TweetSentimentExtractionClassification (default) (test; main_score)
- 58.362Classification
- MTEB TwentyNewsgroupsClustering (default) (test; main_score)
- 38.620Clustering
- MTEB TwitterSemEval2015 (default) (test; main_score)
- 64.413PairClassification
- MTEB TwitterURLCorpus (default) (test; main_score)
- 86.155PairClassification
- Paramters (M)
- 278
- Miracl (18)
- 58.3
- Mintaka Retrieval (8)
- 23.2
- MTEB English (15)
- 48.2
- MTEB German (4)
- 71.2
- MTEB Spanish (2)
- 52.6
- MTEB French (5)
- 54.1
- MTEB Japanese (2)
- 61.7
- MTEB Arabic (1)
- 64.2
- MTEB Korean (1)
- 71.8
- MTEB Chinese (8)
- 45.2
Values as printed on the card, with the publisher's own settings; column “granite-embedding-278M-multilingual”. Not comparable across publishers and not used in any ranking. Model card
Quality × memory
No comparable quality × memory pair
Embedding memory estimates cluster at a few GB, so a quality × memory chart would not separate these models; MTEB is ranked above instead.
Consider Granite Embedding 278M Multilingual for retrieval and semantic search: it turns text into 768-dimension vectors.
The publisher states no memory figure; the hardware list uses a ~3 GB planning estimate, a ceiling and not a measurement. PT-BR is confirmed by a source.
Runs on
Fast memory holds it 119
The planning estimate fits the memory the model can use quickly: graphics card VRAM, or unified memory within the macOS GPU share.
Show all 119
Hardware profiles grouped by the embedding planning estimate (8-bit weights, 20% allowance, 2 GB reserve). Not a measured fit.
Measured speed
No published measurement for this checkpoint
Throughput (texts per second) is shown only when a source publishes it with hardware, build and setup. It is never estimated.
How to run
The planning estimate uses 8-bit weights, 20% working allowance and a 2 GB reserve; it is a ceiling, not a measurement.
- Runs in
- Ollama, Transformerssource
- Languages · PT-BR
- 12 languages: English, German, Spanish, French, Japanese, Portuguese, Arabic, Czech, Italian, Korean, Dutch, Chinese (model card)PT-BR confirmed by a source
Documented ways to run it
Copied as printed in the linked source (read 2026-10-18); not run by this site.
pip install sentence_transformersfrom sentence_transformers import SentenceTransformer, util
model_path = "ibm-granite/granite-embedding-278m-multilingual"
# Load the Sentence Transformer model
model = SentenceTransformer(model_path)
input_queries = [
' Who made the song My achy breaky heart? ',
'summit define'
]
input_passages = [
"Achy Breaky Heart is a country song written by Don Von Tress. Originally titled Don't Tell My Heart and performed by The Marcy Brothers in 1991. ",
"Definition of summit for English Language Learners. : 1 the highest point of a mountain : the top of a mountain. : 2 the highest level. : 3 a meeting or series of meetings between the leaders of two or more governments."
]
# encode queries and passages
query_embeddings = model.encode(input_queries)
passage_embeddings = model.encode(input_passages)
# calculate cosine similarity
print(util.cos_sim(query_embeddings, passage_embeddings))pip install transformers torchimport torch
from transformers import AutoModel, AutoTokenizer
model_path = "ibm-granite/granite-embedding-278m-multilingual"
# Load the model and tokenizer
model = AutoModel.from_pretrained(model_path)
tokenizer = AutoTokenizer.from_pretrained(model_path)
model.eval()
input_queries = [
' Who made the song My achy breaky heart? ',
'summit define'
]
# tokenize inputs
tokenized_queries = tokenizer(input_queries, padding=True, truncation=True, return_tensors='pt')
# encode queries
with torch.no_grad():
# Queries
model_output = model(**tokenized_queries)
# Perform pooling. granite-embedding-278m-multilingual uses CLS Pooling
query_embeddings = model_output[0][:, 0]
# normalize the embeddings
query_embeddings = torch.nn.functional.normalize(query_embeddings, dim=1)
Technical details
What the source and the catalog store for this checkpoint. Missing fields are left out, never zero.
- Checkpoint / source
- ibm-granite/granite-embedding-278m-multilingual
- Task
- embeddings for search (feature extraction)
- Library in source
- Ollama, Transformers
- Weight license
- Apache 2.0
- Commercial use
- Allowed Apache License 2.0 text (SPDX) · read 2026-10-18license text
- Parameter basis
- Hub safetensors total (278,043,648 parameters, read 2026-10-07 from the Hub API); publishers round it in their names.
- Announced
- 2024-12-18
Superseded by Granite Embedding 311M Multilingual R2 (listed separately). Ollama's “granite-embedding” page serves this 278M model and a 30M English one.
Sources
Identity and published specifications come from the checkpoint page. Benchmark scores belong to the named release and do not measure speed on your computer.
- Publisher model card
- Hugging Face model page / card · 2026-10-07
- Hugging Face API (parameters, licence, library) · 2026-10-07
- Ollama library (availability only) · 2026-10-07
- MTEB results dataset (CC0), scores computed by scripts/ingest-mteb-results.mjs · 2026-10-25
Curated facts checked: 2026-10-07. Each value cites its source (the publisher's card or the MTEB results dataset); estimates are labelled as estimates, and a value no source states stays empty. A repository commit is not a model release date.






























































