Embedding APIs
An embedding API turns text into a vector, so you can search by meaning instead of keywords. It is the first step of most RAG pipelines: embed your documents once, store the vectors, then embed each query and look for the nearest match. Prices are cents per million tokens, so the choice comes down to retrieval quality on your own data, context length, languages, and whether images or code go in the same index. Changing models later means re-embedding everything, so test on your documents first. OpenAI, Cohere, Voyage, Google, Jina, Mistral and Amazon train their own models. DeepInfra and Fireworks host open ones. 9 tools, in the order they appear in the directory. This is not a ranking.
Grouped by job: AI Agent Infrastructure
Fireworks AI
Inference platform for open models: per-token serverless with Standard, Priority and Fast modes, batch at half the serverless price, on-demand GPUs billed per second, and LoRA fine-tuning with tuned models served at the base price. Serverless embedding models too, from Qwen3 Embedding 8B to small BERT-size models. OpenAI-compatible.
DeepInfra
OpenAI-compatible API for open models such as Llama, DeepSeek and Qwen, billed per token with Standard, Priority and Flex tiers. Also open text-to-speech models such as Kokoro, billed per character, open embedding models such as Qwen3-Embedding and bge-m3, and dedicated GPUs for your own LLMs billed by the minute.
OpenAI Embeddings
OpenAI's embeddings endpoint: text-embedding-3-small returns 1,536 dimensions and text-embedding-3-large 3,072, and both can be shortened with the dimensions parameter. text-embedding-ada-002 is the older model. Billed per token on your OpenAI key.
Cohere Embed
Cohere's embedding API: Embed v5 Pro and Fast share one embedding space, so you can index with Pro and query with Fast. Text, images and PDFs in 100+ languages, 128K context, and vectors from 256 to 2,048 dimensions as float, int8 or binary. Free trial keys, rate limited.
Voyage AI
Embedding models for retrieval: voyage-4-large, voyage-4 and voyage-4-lite for text, plus code, contextual-chunk and multimodal models. The first 200M tokens are free and batch is 33% off. voyage-4-nano has open weights under Apache-2.0.
Gemini Embedding
Google's embedding models on the Gemini API: gemini-embedding-2 maps text, images, video, audio and documents into one embedding space, and gemini-embedding-001 handles text, with a cheaper batch price. Free tier for testing.
Jina AI Embeddings
Jina's embedding API: jina-embeddings-v5 text and omni models in small (32K context) and nano (8K) sizes, plus code embedding models, behind an OpenAI-compatible endpoint. Weights are on Hugging Face under CC BY-NC 4.0; commercial self-hosting is licensed through Elastic.
Mistral Embed
Mistral's embedding models on the Mistral API: mistral-embed for text with 8K context, and codestral-embed for code. Cached input costs a tenth of the normal price, and codestral-embed is half price in batch.
Amazon Titan Text Embeddings
Amazon's text embedding models on Bedrock: Titan Text Embeddings V2 takes up to 8,192 tokens and returns 1,024, 512 or 256 dimensions, with batch at half price. Nova Multimodal Embeddings sits in the same account for more than text.
What it is, which models, what a million tokens costs.
Straight from each vendor's own pages. Plans move often, so the check date is below.
| Tool | What it is | Models | Price per 1M tokens | Open weights |
|---|---|---|---|---|
| Fireworks AI | Serverless open embedding models on Fireworks' OpenAI-compatible API. Voyage 4 models on dedicated deployments. | qwen3-embedding-8b (40K context), plus BERT-size models such as bge, gte, nomic-embed-text and mxbai-embed-large. | qwen3-embedding-8b $0.10. BERT-size models $0.008 up to 150M parameters, $0.016 from 150M to 350M. | Yes, the models. Fireworks is not. |
| DeepInfra | Hosted open embedding models on DeepInfra's OpenAI-compatible API. | Qwen3-Embedding 0.6B, 4B and 8B, bge-m3, embeddinggemma-300m, Nemotron-3-Embed-1B, e5-large-v2, all-MiniLM-L6-v2 and more. | Qwen3-Embedding 8B and 0.6B $0.01, 4B $0.02. bge-m3 $0.01, embeddinggemma-300m $0.002, Nemotron-3-Embed-1B $0.015, all-MiniLM-L6-v2 $0.005. | Yes, the models. DeepInfra is not. |
| OpenAI Embeddings | OpenAI's embeddings endpoint, on the same key as its other models. Shorten vectors with the dimensions parameter. | text-embedding-3-small (1,536 dimensions), text-embedding-3-large (3,072), text-embedding-ada-002 (older). | 3-small $0.02, 3-large $0.13, ada-002 $0.10. | No. |
| Cohere Embed | Embed v5 for text, images and PDFs in 100+ languages, 128K context. Pro and Fast share one embedding space. | embed-v5.0-pro, embed-v5.0-fast. 256 to 2,048 dimensions, as float, int8 or binary. | Pro $0.12, Fast $0.08. Free trial keys, rate limited. | No. |
| Voyage AI | Retrieval embeddings, with code, contextual-chunk and multimodal models next to the general ones. | voyage-4-large, voyage-4, voyage-4-lite, voyage-code-4, voyage-context-4, voyage-multimodal-3.5, voyage-4-nano. | voyage-4-large $0.12, voyage-4 $0.06, voyage-4-lite $0.02, code and context $0.12, multimodal $0.12 plus $0.60 per billion pixels. First 200M tokens free; batch 33% off. | voyage-4-nano only, Apache-2.0 on Hugging Face. |
| Gemini Embedding | Google's embedding models on the Gemini API. gemini-embedding-2 puts text, images, video, audio and documents in one space. | gemini-embedding-2, gemini-embedding-001. | gemini-embedding-2 text $0.20. gemini-embedding-001 $0.15, batch $0.075. Free tier. | No. |
| Jina AI Embeddings | Text, multimodal and code embeddings behind an OpenAI-compatible API, with free tokens on a new key. | jina-embeddings-v5 text and omni, small (32K context) and nano (8K); jina-code-embeddings 0.5b and 1.5b. | v5 small $0.05, v5 nano $0.02, code $0.05. | Yes, CC BY-NC 4.0, non-commercial. Commercial self-hosting licensed through Elastic. |
| Mistral Embed | Text and code embeddings on the Mistral API. | mistral-embed (8K context), codestral-embed-2505. | mistral-embed $0.10, codestral-embed $0.15. Cached input a tenth of that; codestral-embed batch 50% off. | No. |
| Amazon Titan Text Embeddings | Amazon's embedding models on Bedrock, billed in your AWS account. | Titan Text Embeddings V2 (8,192 tokens; 1,024, 512 or 256 dimensions), Titan Text Embeddings v1, Nova Multimodal Embeddings. | Titan V2 $0.02, batch $0.01. Titan v1 $0.10. Nova Multimodal text $0.135, batch $0.0675. | No. |
Checked 2026-11-02. Sources: OpenAI pricing, OpenAI embeddings guide, Cohere pricing, Cohere Embed docs, Voyage AI pricing, voyage-4-nano on Hugging Face, Gemini API pricing, Gemini embeddings docs, Jina embeddings, Jina models and prices, Mistral pricing, Mistral embeddings docs, Amazon Bedrock pricing, Titan Text Embeddings docs, DeepInfra embedding models, Fireworks pricing, Fireworks embeddings docs.
Pick it if.
- OpenAI EmbeddingsYou already call OpenAI and want embeddings on the same key, with 3-small at $0.02 per million tokens.
- Cohere EmbedYou embed PDFs and images as well as text, or want to index with one model and query with a cheaper one.
- Voyage AIRetrieval is the whole job, you want code and long-document models, and 200M free tokens to test on.
- Gemini EmbeddingYou need images, video or audio in the same index as text, or you build on Google already.
- Jina AI EmbeddingsYou want small, cheap text and multimodal models, and non-commercial weights or an Elastic license work for you.
- Mistral EmbedYou use Mistral's models already, or want a code embedding model on the same API.
- Amazon Titan Text EmbeddingsYou are on AWS and want embeddings billed through Bedrock, at $0.02 per million tokens for Titan V2.
- DeepInfraPrice matters most, and you want open models you could later run yourself.
- Fireworks AIYou run open models on Fireworks already and want embeddings on the same bill.
Each listing.
The directory page, alternatives in the same category, and a head-to-head where we have one.
- Fireworks AIAlternatives
- DeepInfraAlternatives
- OpenAI EmbeddingsAlternatives
- Cohere EmbedAlternatives
- Voyage AIAlternatives
- Gemini EmbeddingAlternatives
- Jina AI EmbeddingsAlternatives
- Mistral EmbedAlternatives
- Amazon Titan Text EmbeddingsAlternatives
Category pages list every tool in that category, including ones left off this page. Built one of these? Get it featured in its category.
The 2 biggest AI stories, every morning.
One short email: the two stories that matter, and why.
You're in.
The Daily AI Brief will land in your inbox soon. If you don't see it, check Promotions or Spam and move it to your main inbox.