Qwen3 Embedding 4B API

Convert text to vectors for search, clustering, and retrieval augmented generation in over 100 languages from $0.009.

2560 dimension vectors
69.45 MTEB multilingual
33K token context
Over 100 languages

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-4B",
    input="The quick brown fox jumps over the lazy dog.",
)

print(response.data[0].embedding)

How to use the Qwen3 Embedding 4B 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 the model to Qwen/Qwen3-Embedding-4B.
3
Call /embeddings with your text in input, put Instruct: <task>\nQuery: <text> in front of a query, and read a vector of 2560 numbers from the reply.

What Qwen3 Embedding 4B is

Qwen3 Embedding 4B maps text to a vector of 2560 numbers, in over 100 written and programming languages, at inputs up to 33K tokens. Three sizes ship in this family, and the 4B is the middle one. It scores near the 8B on MTEB multilingual and returns a shorter vector, so an index of the same corpus costs less to hold. The 0.6B is shorter still and scores lower on that benchmark, so pick it when index size rules.

What it does well

Builds search indexes for retrieval augmented generation apps.
Matches questions in one language to documents in another.
Finds functions in a code base from plain text descriptions.

What it is not for

It returns vectors, not text. It cannot answer questions or explain text matches. Pair it with a reranker or chat model for those tasks.
It reads text only. Images and audio need another model.

Calling it

Post to /embeddings with the model string Qwen/Qwen3-Embedding-4B. Pass one string in input, or an array of strings for a batch that returns the vectors in the same order. The dimensions field cuts the vector below 2560 numbers, down to 32, which saves index memory and loses some accuracy. Callers miss the instruction prefix: put Instruct: <task>\nQuery: <text> in front of a query, leave the stored documents plain, and write the instruction in English. Qwen measures a drop of one to five percent on retrieval when the query carries no instruction.

Qwen/Qwen3-Embedding-4B /embeddings

SDKs and languages

Qwen3 Embedding 4B 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	4B
Active parameters	dense, all of them
Score	69.45 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	2560 dimensions
Pulls	2M in the last 30 daysWeights pulled from Hugging Face, read on 2026-09-24. It counts the repository, not this API.
License	Apache 2.0

Price

Input

$0.009

per million tokens

Output

n/a

this model returns a vector

Against OpenRouter

10%

below the cheapest of 2 providers

The cheapest OpenRouter price for this model was $0.010 in and n/a out, with a median of $0.010 and n/a across the comparable providers. Read on 2026-09-17.

License

Qwen releases the model under Apache 2.0. Commercial use is allowed without extra vendor terms.

Parameters: 7.953B
Context: 33K
Vector: 4096
Input per 1M: $0.031
Output per 1M: n/a
Released: Jul 2026
Pulls, 30 days: 97.1K
License: OpenMDW 1.1

Qwen3 Embedding 8BQwen/Qwen3-Embedding-8B 🤗

Parameters: 7.6B
Context: 33K
Vector: 4096
Input per 1M: $0.009
Output per 1M: n/a
Released: Jun 2025
Pulls, 30 days: 2.6M
License: Apache 2.0

Qwen3 Embedding 0.6BQwen/Qwen3-Embedding-0.6B 🤗

Parameters: 0.6B
Context: 33K
Vector: 1024
Input per 1M: $0.004
Output per 1M: n/a
Released: Jun 2025
Pulls, 30 days: 9.1M
License: Apache 2.0

Start using Qwen3 Embedding 4B

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

Get API key Read the quickstart