Pinecone model gallery - Pinecone Docs

Models

Compare the embedding and reranking models you can use with Pinecone.

Pinecone API key
Bring your own vectors

Embed and rerank with these models using only your Pinecone API key. Let Pinecone embed your text and rerank results as part of upserts and searches, or call a model directly to get vectors or scores. The Supported API column shows which options each model supports.

Model	Company	Type	Modality	Dimension	Max tokens	Processing location	Supported API	Actions
llama-text-embed-v2	NVIDIA	Embedding	Text	1024, 2048, 768, 512, 384	2048	`us`	Documents, Standalone embedding, Records	Create index
multilingual-e5-large	Microsoft	Embedding	Text	1024	507	`us`	Documents, Standalone embedding, Records	Create index
pinecone-sparse-english-v0	Pinecone	Embedding	Text	Sparse	2048	`us`	Documents, Standalone embedding, Records	Create index
cohere-rerank-4-fast	Cohere	Reranking	Text	N/A	4096	`global`	Standalone reranking, Records
bge-reranker-v2-m3	BAAI	Reranking	Text	N/A	1024	`us`	Standalone reranking, Records

In the Supported API column:

Documents: The Documents API, for document indexes with integrated embedding.
Standalone embedding: The standalone /embed endpoint, which embeds text without an index.
Standalone reranking: The standalone /rerank endpoint, which reranks results without an index.
Records: The legacy Records API, for legacy integrated indexes created with create_index_for_model.

These models don’t work with a Pinecone API key. Generate vectors with the company’s API, or run an open-source model yourself, and upsert them to an index. To rerank, call the company’s API with the results you query from Pinecone.

Model	Company	Type	Modality	Dimension	Max tokens	Actions
gte-base	Alibaba	Embedding	Text	768	512	Create index
gte-large	Alibaba	Embedding	Text	1024	512	Create index
embed-english-light-v3.0	Cohere	Embedding	Text	384	512	Create index
embed-english-v3.0	Cohere	Embedding	Text	1024	512	Create index
embed-multilingual-v3.0	Cohere	Embedding	Text	1024	512	Create index
all-MiniLM-L12-v2	HuggingFace	Embedding	Text	384	256	Create index
all-mpnet-base-v2	HuggingFace	Embedding	Text	768	384	Create index
instructor-large	HuggingFace	Embedding	Text	768	512	Create index
instructor-xl	HuggingFace	Embedding	Text	768	512	Create index
jina-clip-v2	Jina AI	Embedding	Multimodal	1024	8192	Create index
jina-embeddings-v4	Jina AI	Embedding	Multimodal	2048, 128, 256, 512, 1024	32768	Create index
CLIP-ViT-B-32-laion2B-s34B-b79K	Laion	Embedding	Multimodal	512	77	Create index
e5-base-v2	Microsoft	Embedding	Text	768	512	Create index
e5-large-v2	Microsoft	Embedding	Text	1024	512	Create index
mistral-embed	Mistral AI	Embedding	Text	1024	8192	Create index
CLIP	OpenAI	Embedding	Multimodal	512, 768	77	Create index
text-embedding-ada-002	OpenAI	Embedding	Text	1536	8192	Create index
rerank-2	Voyage AI	Reranking	Text	N/A	16000
rerank-2-lite	Voyage AI	Reranking	Text	N/A	8000
voyage-2	Voyage AI	Embedding	Text	1024	4000	Create index
voyage-3	Voyage AI	Embedding	Text	1024	32000	Create index
voyage-3-large	Voyage AI	Embedding	Text	1024, 256, 512, 2048	32000	Create index
voyage-3-lite	Voyage AI	Embedding	Text	512	32000	Create index
voyage-code-2	Voyage AI	Embedding	Text	1536	16000	Create index
voyage-code-3	Voyage AI	Embedding	Text	1024, 256, 512, 2048	32000	Create index
voyage-finance-2	Voyage AI	Embedding	Text	1024	32000	Create index
voyage-large-2	Voyage AI	Embedding	Text	1536	16000	Create index
voyage-law-2	Voyage AI	Embedding	Text	1024	16000	Create index
voyage-multimodal-3	Voyage AI	Embedding	Multimodal	1024	32000	Create index