Embedding models - verified
Embedding models compared
59 embedding models from 14 providers: output dimensions, max input tokens, what each one can embed and the first-party list price per 1M tokens, every value linked to the provider's own page.
- The lowest list price among current API models is $0.02 per 1M input tokens, shared by text-embedding-3-small (OpenAI), Voyage 4 Lite (Voyage AI) and Amazon Titan Text Embeddings V2 (Amazon).
- The longest stated input is 128,000 tokens (Embed 5 Pro, Embed 5 Fast, Embed 4, Qwen3.7 Text Embedding and Qwen3.7 Text Embedding Flash).
- The largest vectors are 4,096 dimensions (Qwen3-Embedding-8B and Nemotron 3.5 Embed 8B).
Want to run one yourself? 10 open-weight embedding models, sorted by size with licenses.
- Models
- 59
- Providers
- 14
- Current
- 45
- Open weights
- 10
Every embedding model
Current models first, legacy ones greyed out below. Flexible means the provider documents shorter output vectors. Prices are standard list prices in USD per 1M input tokens, with the batch rate underneath when one is listed.
59 models
| Model | Provider | Access | Status | Default dims | Max input | Inputs | Price / 1M tokensbatch below | License |
|---|---|---|---|---|---|---|---|---|
| No models match these filters. | ||||||||
text-embedding-3-largetext-embedding-3-large | OpenAI | API | Current | 3,072Flexible | 8,192 | Text | $0.13 | -API only, no open-weight license |
text-embedding-3-smalltext-embedding-3-small | OpenAI | API | Current | 1,536Flexible | 8,192 | Text | $0.02 | -API only, no open-weight license |
Gemini Embedding 2gemini-embedding-2 | API | Current | 3,072Flexible | 8,192 | Text, Image, Audio, Video, PDF | $0.20batch $0.10 | -API only, no open-weight license | |
text-embedding-005text-embedding-005 | API | Current | 768Flexible | 2,048 | Text | Not stated | -API only, no open-weight license | |
text-multilingual-embedding-002text-multilingual-embedding-002 | API | Current | 768Flexible | 2,048 | Text | Not stated | -API only, no open-weight license | |
Multimodal Embeddings (multimodalembedding@001)multimodalembedding@001 | API | Current | 1,408Flexible | 32 | Text, Image, Video | Not stated | -API only, no open-weight license | |
Embed 5 Proembed-v5.0-pro | Cohere | API | Current | 2,048Flexible | 128,000 | Text, Image, PDF | $0.12 | -API only, no open-weight license |
Embed 5 Fastembed-v5.0-fast | Cohere | API | Current | 2,048Flexible | 128,000 | Text, Image, PDF | $0.08 | -API only, no open-weight license |
Embed 4embed-v4.0 | Cohere | API | Current | 1,536Flexible | 128,000 | Text, Image, PDF | Not stated | -API only, no open-weight license |
Embed English v3.0embed-english-v3.0 | Cohere | API | Current | 1,024 | 512 | Text, Image | Not stated | -API only, no open-weight license |
Embed English Light v3.0embed-english-light-v3.0 | Cohere | API | Current | 384 | 512 | Text, Image | Not stated | -API only, no open-weight license |
Embed Multilingual v3.0embed-multilingual-v3.0 | Cohere | API | Current | 1,024 | 512 | Text, Image | Not stated | -API only, no open-weight license |
Embed Multilingual Light v3.0embed-multilingual-light-v3.0 | Cohere | API | Current | 384 | 512 | Text, Image | Not stated | -API only, no open-weight license |
Mistral Embedmistral-embed | Mistral AI | API | Current | 1,024 | 8,000 | Text | $0.10 | -API only, no open-weight license |
Codestral Embedcodestral-embed | Mistral AI | API | Current | 1,536Flexible | 8,000 | Text, Code | $0.15 | -API only, no open-weight license |
Voyage 4 Largevoyage-4-large | Voyage AI | API | Current | 1,024Flexible | 32,000 | Text | $0.12 | -API only, no open-weight license |
Voyage 4voyage-4 | Voyage AI | API | Current | 1,024Flexible | 32,000 | Text | $0.06 | -API only, no open-weight license |
Voyage 4 Litevoyage-4-lite | Voyage AI | API | Current | 1,024Flexible | 32,000 | Text | $0.02 | -API only, no open-weight license |
Voyage Code 4voyage-code-4 | Voyage AI | API | Current | 1,024Flexible | 32,000 | Text, Code | $0.12 | -API only, no open-weight license |
Voyage Context 4voyage-context-4 | Voyage AI | API | Current | 1,024Flexible | 32,000 | Text | $0.12 | -API only, no open-weight license |
Voyage Multimodal 3.5voyage-multimodal-3.5 | Voyage AI | API | Current | 1,024Flexible | 32,000 | Text, Image, Video | $0.12 | -API only, no open-weight license |
Voyage Finance 2voyage-finance-2 | Voyage AI | API | Current | 1,024 | 32,000 | Text | $0.12 | -API only, no open-weight license |
Voyage Law 2voyage-law-2 | Voyage AI | API | Current | 1,024 | 16,000 | Text | $0.12 | -API only, no open-weight license |
Voyage 4 Nanovoyageai/voyage-4-nano | Voyage AI | Open weights | Current | 1,024Flexible | 32,000 | Text | Self-hosted (open weights, no API price) | apache-2.0Commercial use allowed |
Amazon Titan Text Embeddings V2amazon.titan-embed-text-v2:0 | Amazon | API | Current | 1,024Flexible | 8,192 | Text | $0.02 | -API only, no open-weight license |
Amazon Titan Embeddings G1 - Textamazon.titan-embed-text-v1 | Amazon | API | Current | Not stated | 8,192 | Text | Not stated | -API only, no open-weight license |
Amazon Titan Multimodal Embeddings G1amazon.titan-embed-image-v1 | Amazon | API | Current | 1,024Flexible | 256 | Text, Image | Not stated | -API only, no open-weight license |
Amazon Nova Multimodal Embeddingsamazon.nova-2-multimodal-embeddings-v1:0 | Amazon | API | Current | Not statedFlexible | 8,000 | Text, Image, Audio, Video | $0.135batch $0.0675 | -API only, no open-weight license |
Qwen3.7 Text Embeddingqwen3.7-text-embedding | Alibaba Cloud | API | Current | 1,024Flexible | 128,000 | Text, Code | $0.07 | -API only, no open-weight license |
Qwen3.7 Text Embedding Flashqwen3.7-text-embedding-flash | Alibaba Cloud | API | Current | Not statedFlexible | 128,000 | Text | Not stated | -API only, no open-weight license |
Text Embedding V4text-embedding-v4 | Alibaba Cloud | API | Current | 1,024Flexible | 8,192 | Text, Code | $0.07 | -API only, no open-weight license |
Text Embedding V3text-embedding-v3 | Alibaba Cloud | API | Current | 1,024Flexible | 8,192 | Text | $0.07 | -API only, no open-weight license |
Tongyi Embedding Vision Plustongyi-embedding-vision-plus | Alibaba Cloud | API | Current | 1,152Flexible | 1,024 | Text, Image, Video | $0.09 | -API only, no open-weight license |
Tongyi Embedding Vision Flashtongyi-embedding-vision-flash | Alibaba Cloud | API | Current | 768Flexible | 1,024 | Text, Image, Video | $0.09 | -API only, no open-weight license |
Qwen3-VL Embeddingqwen3-vl-embedding | Alibaba Cloud | API | Current | 2,560Flexible | 32,000 | Text, Image, Video | $0.10 | -API only, no open-weight license |
Multimodal Embedding V1multimodal-embedding-v1 | Alibaba Cloud | API | Current | 1,024 | 512 | Text, Image, Video | Not stated | -API only, no open-weight license |
Qwen3-Embedding-0.6BQwen/Qwen3-Embedding-0.6B | Qwen (Alibaba) | Open weights | Current | 1,024Flexible | 32,768 | Text, Code | Self-hosted (open weights, no API price) | apache-2.0Commercial use allowed |
Qwen3-Embedding-4BQwen/Qwen3-Embedding-4B | Qwen (Alibaba) | Open weights | Current | 2,560Flexible | 32,768 | Text, Code | Self-hosted (open weights, no API price) | apache-2.0Commercial use allowed |
Qwen3-Embedding-8BQwen/Qwen3-Embedding-8B | Qwen (Alibaba) | Open weights | Current | 4,096Flexible | 32,768 | Text, Code | Self-hosted (open weights, no API price) | apache-2.0Commercial use allowed |
EmbeddingGemma 2google/embeddinggemma-2 | Google DeepMind | Open weights | Current | 768Flexible | 8,192 | Text, Code, Image, Audio, Video | Self-hosted (open weights, no API price) | apache-2.0Commercial use allowed |
BGE-M3BAAI/bge-m3 | BAAI | Open weights | Current | 1,024 | 8,192 | Text | Self-hosted (open weights, no API price) | mitCommercial use allowed |
Nomic Embed Text v2 MoEnomic-ai/nomic-embed-text-v2-moe | Nomic AI | Open weights | Current | 768Flexible | 512 | Text | Self-hosted (open weights, no API price) | apache-2.0Commercial use allowed |
Granite Embedding 311M Multilingual R2ibm-granite/granite-embedding-311m-multilingual-r2 | IBM | Open weights | Current | 768Flexible | 32,768 | Text, Code | Self-hosted (open weights, no API price) | apache-2.0Commercial use allowed |
Nemotron 3.5 Embed 8Bnvidia/Nemotron-3.5-Embed-8B-BF16 | NVIDIA | Open weights | Current | 4,096Flexible | 32,768 | Text, Image | Self-hosted (open weights, no API price) | openmdw-1.1Commercial use allowed |
GTE ModernBERT BaseAlibaba-NLP/gte-modernbert-base | Alibaba (Tongyi Lab) | Open weights | Current | 768 | 8,192 | Text | Self-hosted (open weights, no API price) | apache-2.0Commercial use allowed |
text-embedding-ada-002text-embedding-ada-002 | OpenAI | API | Legacy | 1,536 | 8,192 | Text | $0.10 | -API only, no open-weight license |
Gemini Embeddinggemini-embedding-001 | API | Legacy | 3,072Flexible | 2,048 | Text | Not stated | -API only, no open-weight license | |
Voyage 3 Largevoyage-3-large | Voyage AI | API | Legacy | 1,024Flexible | 32,000 | Text | $0.18 | -API only, no open-weight license |
Voyage 3.5voyage-3.5 | Voyage AI | API | Legacy | 1,024Flexible | 32,000 | Text | $0.06 | -API only, no open-weight license |
Voyage 3.5 Litevoyage-3.5-lite | Voyage AI | API | Legacy | 1,024Flexible | 32,000 | Text | $0.02 | -API only, no open-weight license |
Voyage Code 3voyage-code-3 | Voyage AI | API | Legacy | 1,024Flexible | 32,000 | Text, Code | $0.18 | -API only, no open-weight license |
Voyage Context 3voyage-context-3 | Voyage AI | API | Legacy | 1,024Flexible | 32,000 | Text | $0.18 | -API only, no open-weight license |
Voyage Multimodal 3voyage-multimodal-3 | Voyage AI | API | Legacy | 1,024 | 32,000 | Text, Image | $0.12 | -API only, no open-weight license |
Voyage 3voyage-3 | Voyage AI | API | Legacy | 1,024 | 32,000 | Text | $0.06 | -API only, no open-weight license |
Voyage 3 Litevoyage-3-lite | Voyage AI | API | Legacy | 512 | 32,000 | Text | $0.02 | -API only, no open-weight license |
Voyage Multilingual 2voyage-multilingual-2 | Voyage AI | API | Legacy | 1,024 | 32,000 | Text | $0.12 | -API only, no open-weight license |
Voyage Large 2 Instructvoyage-large-2-instruct | Voyage AI | API | Legacy | 1,024 | 16,000 | Text | $0.12 | -API only, no open-weight license |
Voyage Large 2voyage-large-2 | Voyage AI | API | Legacy | 1,536 | 16,000 | Text | $0.12 | -API only, no open-weight license |
Voyage 2voyage-2 | Voyage AI | API | Legacy | 1,024 | 4,000 | Text | $0.10 | -API only, no open-weight license |
Method
How to read these numbers
Read from official provider docs, pricing pages and Hugging Face model cards on . Prices and limits change; check the linked source before you commit to re-embedding a corpus.
- First-party list prices
- Prices are the provider's own list price per 1M input tokens at the standard rate, with the batch rate when one is listed, copied from its pricing page. Discounts, free tiers and cloud marketplace prices are not included.
- Not a quality benchmark
- Nothing here measures retrieval quality or speed. Tables sort by price, input length or dimensions only. Shortlist on these facts, then evaluate on your own queries and documents.
- Not stated means not stated
- When the provider's docs do not give a value, the cell says so instead of guessing. Some providers price per character, image or request; those prices stay in the note on the model page.
- Snippets copied verbatim
- Usage snippets come from the page linked under each one: the provider's docs or the Hugging Face model card. Models whose sources publish no snippet show none.
FAQ
What is the cheapest embedding model API?
The lowest list price among current API models is $0.02 per 1M input tokens, shared by text-embedding-3-small (OpenAI), Voyage 4 Lite (Voyage AI) and Amazon Titan Text Embeddings V2 (Amazon). 36 of 59 models have a per-token list price on the provider's own pricing page; 2 list a separate batch rate. A low price says nothing about retrieval quality, so test on your own data before you switch.
Which embedding models accept the longest input?
Embed 5 Pro, Embed 5 Fast, Embed 4, Qwen3.7 Text Embedding and Qwen3.7 Text Embedding Flash accept up to 128,000 tokens per input. The median stated max input across the directory is 16,000 tokens and the shortest is 32, so check the max input column before you pick a chunk size.
Which embedding models output the most dimensions?
Qwen3-Embedding-8B and Nemotron 3.5 Embed 8B can return 4,096-dimension vectors. Bigger vectors cost more to store and search. 40 of 59 models document flexible output sizes (Matryoshka-style), so you can ask for shorter vectors instead of truncating them yourself.
Which embedding models can embed images, audio or video?
18 current models accept more than text and code: Gemini Embedding 2, Multimodal Embeddings (multimodalembedding@001), Embed 5 Pro, Embed 5 Fast, Embed 4, Embed English v3.0, Embed English Light v3.0, Embed Multilingual v3.0 and 10 more. The inputs column lists exactly what each one takes, and each model page links the provider's limits.
Which embedding models can I run myself?
10 models publish open weights on Hugging Face, and 10 of them ship under licenses that allow commercial use. The open-weight list, sorted by size with each license, is at https://www.developersdigest.tech/models/embeddings/open-weights.
Why are some models marked legacy?
14 of 59 models are older generations their providers still list, usually beside a newer model they point new work to. They stay in the table, greyed out, because existing indexes depend on them: you cannot mix vectors from two models in one index, so switching means re-embedding everything.
Is this a benchmark or a ranking?
No. Every number here is a first-party fact: output dimensions, max input tokens, input types and list prices, copied from provider docs, pricing pages and Hugging Face model cards. Tables sort by those facts only. Nothing measures retrieval quality, so treat price and size as filters, then evaluate the shortlist on your own data.
What does Not stated mean?
The provider's own pages do not give the value, so this directory leaves it blank rather than guessing. Some prices are listed per character, per image or per request instead of per token; those show the provider's note on the model page.
Newsletter
New models, every week
One email per week. New model releases, embedding and retrieval updates and what they mean for what you build.
Keep exploring
Model API pricing and specs
First-party prices per 1M tokens for chat and reasoning models.
Local models
Open-weight LLMs you can download and run, with sizes and licenses.
What is RAG?
Where embeddings fit in a retrieval pipeline, end to end.
Vector databases compared
Where the vectors go once you have picked a model.