DEVELOPER REFERENCE · VERIFIED MODEL EVIDENCE

Compare 13 source-verified model records, repository licenses, published weight formats and dated evidence. No overall model ranking.

Evidence statusSource-first verified
Latest record check2026-11-13
Latest record full verification2026-10-09

DIRECT ANSWER · SOURCE-OBSERVED REGISTRY

What Sentence Transformers model evidence is available?

Open Model Weights currently tracks 13 field-verified Sentence Transformers model records. In this indexed sample, the recorded modalities are 13 Text embeddings. The published repository license labels include 11 apache-2.0; 2 Unknown / not declared. The listed weight formats include ONNX (13), PyTorch (13), Safetensors (13).

Coverage describes the model repositories indexed here, not every model released by the publisher. A checked source, an unknown field and an unchanged source revision are different states. No missing value is treated as a negative answer.

13verified model records
1recorded modalities
2distinct license labels
13model repository links

SOURCE-BACKED PUBLISHER PROFILE

Sentence Transformers: models, research and access

Sentence Transformers is an open-source ecosystem and model collection for creating dense text embeddings and semantic similarity scores. Its official pretrained-model documentation distinguishes general sentence embedding, multilingual and question-answering retrieval checkpoint families.

These are embedding encoders, not general-purpose chat assistants. Model output dimensions, tokenizer and language scope, vector normalization, similarity metric and query/passage setup determine whether two checkpoints can be substituted inside a search index.

01

Why embeddings are different from text generation

A Sentence Transformer maps sentences and passages into numerical vectors for semantic similarity, clustering and retrieval. The official all-MiniLM-L6-v2 card describes 384-dimensional embeddings. Embedding models do not themselves generate a prose response: in RAG they supply retrieved passages to a separate model.

02

MiniLM, MPNet and quality/footprint trade-offs

The pretrained-model documentation lists MiniLM and MPNet as separate encoder families. Their dimensionality, sequence handling and computational footprints differ; one model's embedding vectors cannot simply replace another model's vectors in an existing index. Re-embed the corpus and evaluate retrieval on representative queries.

03

Multilingual and question-answering families

Multilingual MiniLM and MPNet models target cross-language semantic representation. Multi-QA models are trained for query-to-passage retrieval with specific dot-product or cosine scoring conventions, while LaBSE supports multilingual sentence representations. Use the documented metric and evaluate language-specific performance.

04

Library license versus individual model license

The Sentence Transformers library documentation covers model loading and training, but software licensing and downloaded model-card licenses are distinct. Most of the 13 indexed checkpoint records have Apache-2.0 labels; two Multi-QA records currently lack a declared license in this registry. Unknown is not a denial or permission.

05

Practical RAG setup and observability

Specify the query/corpus instructions, embedding size, similarity function and truncation policy when indexing. Source evidence does not establish customer-specific recall, latency or GPU requirements. Compare embedding retrieval and any second-stage reranker as independent pipeline components.

Publisher background is attributed to official sources. Model counts and license classifications come from OMW repository evidence; no model benchmark or processing location is inferred.

MODEL-BY-MODEL REFERENCE

Which Sentence Transformers embedding should you inspect?

Browse by published modality and licensing evidence. These are navigation cues, not quality, speed or suitability rankings.

Text embeddingsapache-2.0

all-MiniLM-L6-v2

Compact semantic embeddings

Review the 384-dimensional MiniLM sentence encoder as a retrieval or clustering baseline.

Commercial use: allowedLicense evidence →
Text embeddingsapache-2.0

all-mpnet-base-v2

General semantic embeddings

Inspect MPNet's vector size and model card before changing an existing corpus embedding.

Commercial use: allowedLicense evidence →
Text embeddingsapache-2.0

LaBSE

Cross-language sentence embeddings

Inspect LaBSE for cross-lingual comparison and its own tokenizer/normalization requirements.

Commercial use: allowedLicense evidence →
Text embeddingsUnknown / not declared

multi-qa-MiniLM-L6-cos-v1

Cosine-scored Multi-QA retrieval

Review task-specific cosine similarity and unresolved license evidence before integrating.

Commercial use: unknownLicense evidence →

FIELD-BY-FIELD COMPARISON · PRIMARY LINKS

Model evidence at a glance.

Compare the exact entries in our registry. Parameter values are taken from the source-checked record; a license label is not a legal opinion, and recognized weight files are not an inference-speed benchmark. Open each record for the underlying citations and observed history.

Model and parameter countModalityLicense evidenceFormats and filesSource and history
all-MiniLM-L6-v20.0227137B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →
paraphrase-multilingual-MiniLM-L12-v20.117654B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →
all-mpnet-base-v20.109487B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →
paraphrase-multilingual-mpnet-base-v20.278044B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →
all-MiniLM-L12-v20.0333605B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →
all-distilroberta-v10.0821189B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →
paraphrase-mpnet-base-v20.109487B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →
distiluse-base-multilingual-cased-v10.134734B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors15 filesRepository ↗Model evidence →History →
paraphrase-MiniLM-L6-v20.0227137B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →
multi-qa-mpnet-base-dot-v10.109487B published parametersText embeddingsUnknown / not declaredunknown commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →
LaBSE0.470927B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors5 filesRepository ↗Model evidence →History →
distiluse-base-multilingual-cased-v20.134734B published parametersText embeddingsapache-2.0allowed commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors15 filesRepository ↗Model evidence →History →
multi-qa-MiniLM-L6-cos-v10.0227137B published parametersText embeddingsUnknown / not declaredunknown commercial-use classificationInspect license evidence →ONNX, PyTorch, Safetensors13 filesRepository ↗Model evidence →History →

LICENSE & COVERAGE · NOT LEGAL ADVICE

How do the recorded licenses differ?

Different models from the same developer can use different license terms. The distribution below describes the labels and field-level classifications observed in the listed repositories; it does not automatically establish permissions for a specific deployment.

License labels

  • apache-2.011 of 13
  • Unknown / not declared2 of 13

Commercial-use classifications

  • allowed11 of 13
  • unknown2 of 13

Model modalities

  • Text embeddings13 of 13

A classification of unknown is not a prohibition. Even a recognized permissive license can carry conditions. Review the license for the exact model and revision. Open the license explorer →

EVIDENCE METHOD · TRANSPARENT LIMITS

How this developer profile is assembled.

We aggregate the current source-checked model records, rather than publish an unverified corporate biography. Repository metadata, model cards, license files and recognized weight artifacts supply the facts where available. The record-check date does not assert that a model or its license changed on that date.

Source-derived runtime signals are not hardware benchmarks. Weight-only memory estimates omit runtime overhead, activations and other deployment requirements. Model-specific uncertainties remain visible in the individual evidence records.

QUESTIONS · DIRECT ANSWERS

Frequently asked questions about Sentence Transformers models.

Which Sentence Transformers models are covered?

Open Model Weights currently lists 13 field-verified records for Sentence Transformers. Their recorded model modalities are: 13 Text embeddings. This is a registry sample, not an exhaustive publisher catalog.

Which licenses are recorded for Sentence Transformers models?

In the 13 indexed records, repository license labels are: 11 apache-2.0; 2 Unknown / not declared. License labels do not replace the model-specific terms or establish rights for every use case.

Where can I verify Sentence Transformers model information?

Each indexed model links to its official source repository and a field-level Open Model Weights profile. The most recent record checks shown here reach 2026-11-13; checking a repository does not imply that its contents changed.

Do Sentence Transformers produce chatbot answers?

No. Standard Sentence Transformer models produce embeddings for similarity, search or clustering. A separate generative language model is normally used to write answers in a RAG workflow.

Can MiniLM and MPNet embeddings share an unchanged vector index?

Not safely by default. Model embedding dimensions, score normalization and representation differ. Re-embed documents with the chosen model and match the corresponding query encoder.

Are all 13 indexed Sentence Transformers models Apache-2.0?

No. Most currently carry Apache-2.0 labels, but two Multi-QA entries have unknown or undeclared license labels in the checked registry. Verify the individual model card and terms.

COMPLETE INDEXED SELECTION

All 13 indexed model profiles.