Qwen3 Embedding 0.6B API

Turn text into vectors for search, clustering and retrieval from $0.004 per million input tokens.

Vector width 1024
MTEB score 64.34
Context 33K tokens
$0.004 per million

from openai import OpenAI

client = OpenAI(
    base_url="https://api.modelsheep.com",
    api_key=API_KEY_MODELSHEEP,
)

response = client.embeddings.create(
    model="Qwen/Qwen3-Embedding-0.6B",
    input="The quick brown fox jumps over the lazy dog.",
)

print(response.data[0].embedding)

How to use the Qwen3 Embedding 0.6B API

1
Create a ModelSheep account and copy an API key from the dashboard.
2
Point the OpenAI SDK at the base url https://api.modelsheep.com and set model to Qwen/Qwen3-Embedding-0.6B.
3
Call /embeddings with your text, prefix each query with a one-sentence instruction to keep retrieval accuracy, and read back a vector of 1024 floats.

What Qwen3 Embedding 0.6B is

Qwen3 Embedding 0.6B is the smallest of the three Qwen3 embedding models. All three build on the Qwen3 foundation models, read up to 33K tokens, and cover more than 100 languages, including programming languages. This model returns a narrower vector than the 4B and the 8B, and it scores lower on MTEB multilingual v2. It costs less per million tokens, and its narrower vector needs less index memory.

What it does well

Builds a semantic search index for text in over 100 languages, including
Embeds millions of text chunks for a retrieval augmented generation (RAG)
Groups tickets, reviews, or log lines by meaning for clustering and

What it is not for

Ranking a shortlist. The model returns vectors, not scores for text pairs.
Work that puts accuracy first. This model scores below Qwen3 Embedding 4B and

Calling it

Call POST /embeddings with model set to Qwen/Qwen3-Embedding-0.6B, and send one string or an array of strings. Set dimensions to shorten the vector, down to 32, which saves index memory and costs some accuracy. Prefix each query with a one-sentence instruction and leave the stored documents plain, because Qwen measures a drop of 1 to 5 percent in retrieval quality without it.

Qwen/Qwen3-Embedding-0.6B /embeddings

SDKs and languages

Qwen3 Embedding 0.6B answers on an HTTP endpoint, so any language that can post JSON reaches it. These are the ones the box above writes out.

OpenAI SDK: PythonTypeScriptGoJavaC#PHPRust
HTTP: cURLPythonTypeScriptGoRustJavaC#PHP

Specifications

Author	Qwen
Category	embedding
Accepts	text
Returns	vector
Total parameters	0.6B
Active parameters	dense, all of them
Score	64.34 on MTEB multilingual v2
Languages	100+One claim for the whole Qwen3 Embedding series. The total counts programming languages, so the natural language count is smaller and unstated.
Precision served	FP8
Context	33K tokens
Vector	1024 dimensions
License	Apache 2.0

Price

Input

$0.004

per million tokens

Output

n/a

this model returns a vector

Against OpenRouter

20%

below the cheapest of 2 providers

The cheapest OpenRouter price for this model was $0.005 in and n/a out, with a median of $0.005 and n/a across the comparable providers. Read on 2026-09-17.

License

Qwen releases the weights under Apache 2.0. Commercial use needs no separate permission, and no vendor terms sit on top of the license.

Parameters: 7.953B
Context: 33K
Vector: 4096
Input per 1M: $0.031
Output per 1M: n/a
License: OpenMDW 1.1

EmbeddingGemma 300Mgoogle/embeddinggemma-300m

Parameters: 0.308B
Context: 2K
Vector: 768
Input per 1M: $0.001
Output per 1M: n/a
License: Gemma

BGE M3BAAI/bge-m3

Parameters: 0.568B
Context: 8K
Vector: 1024
Input per 1M: $0.009
Output per 1M: n/a
License: MIT

Start using Qwen3 Embedding 0.6B

Point the OpenAI SDK at api.modelsheep.com. Your prompts stay in Germany, and nothing is retained.

Get API key Read the quickstart