Models
DeepSeek API: V4 Flash and V4 Pro pricing
Call DeepSeek V4.1 Flash and V4 Pro through an OpenAI-compatible API. Per-token prices, cache-read rates, tools, JSON and streaming from one Tokely key.
tokely.me/deepseek-api
Embeddings API at half of OpenAI's price
OpenAI text-embedding-3-small, text-embedding-3-large and ada-002 through an OpenAI-compatible /v1/embeddings endpoint, at 50% of OpenAI's list price.
By TokelyReviewed Tokely serves OpenAI's text-embedding-3-small, text-embedding-3-large and text-embedding-ada-002 at /v1/embeddings with the same request and response shape as OpenAI. At launch they cost $0.01, $0.065 and $0.05 per million input tokens, half of OpenAI's list price, from the same key and balance as chat models.
Prices
| Model | Tokely | OpenAI | Saving | Dimensions |
|---|---|---|---|---|
text-embedding-3-small | $0.010 | $0.020 | 50% | 1536, shorter via dimensions |
text-embedding-3-large | $0.065 | $0.130 | 50% | 3072, shorter via dimensions |
text-embedding-ada-002 | $0.050 | $0.100 | 50% | 1536 |
Connect in one request
from openai import OpenAI
import os
client = OpenAI(base_url="https://api.tokely.me/v1", api_key=os.environ["TOKELY_API_KEY"])
chunks = ["Tokely bills per token.", "One key for every model."]
vectors = client.embeddings.create(model="text-embedding-3-small", input=chunks, dimensions=512)
print(len(vectors.data), len(vectors.data[0].embedding))Base URL https://api.tokely.me/v1. Create a key in the dashboard and keep it in an environment variable. Quickstart · Model IDs and endpoints
Built for RAG pipelines
Embed documents and queries with the same model, store the vectors in any vector database, and generate answers with any chat model on the same key. The endpoint accepts batches of up to 2048 inputs and the dimensions parameter for shorter text-embedding-3 vectors.
Drop-in for the OpenAI SDK
Change base_url and the key; client.embeddings.create, LangChain's OpenAIEmbeddings and LlamaIndex's OpenAIEmbedding work unchanged. Vectors are the same models' outputs, so existing indexes stay compatible when the model ID is unchanged.
Billing
Embeddings are billed per input token with no output charge. Current rates are listed on the pricing page and in the dashboard; each request's tokens and charge appear in its log.
Questions before you switch
Are these the same vectors as OpenAI's API?
They come from the same OpenAI embedding models, so vectors are compatible with indexes built with the same model ID and dimensions.
Which model should I use?
text-embedding-3-small for most RAG and search workloads; text-embedding-3-large when retrieval quality matters more than cost; ada-002 only for existing indexes built with it.
START WITH YOUR OWN WORKLOADKnow the price.
Know the price.
Then make the call.
Check the model, compare the rate, and connect with a Tokely key.