Embeddings

TensorX serves text embeddings through the standard OpenAI-compatible /v1/embeddings endpoint. One embedding model is available:

Model	Output dimensions	Max input	Pricing
qwen/qwen3-embedding-8b	4096 (fixed)	40,960 tokens	See tensorx.ai/models

Basic usage

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://api.tensorx.ai/v1"
)

resp = client.embeddings.create(
    model="qwen/qwen3-embedding-8b",
    input="The capital of France is Paris."
)

vector = resp.data[0].embedding   # list of 4096 floats

Batch inputs by passing a list. Each item gets its own vector, returned in the same order:

resp = client.embeddings.create(
    model="qwen/qwen3-embedding-8b",
    input=["first document", "second document", "third document"]
)
vectors = [d.embedding for d in resp.data]

Input limit

The model context length is 40,960 tokens. Requests above this are rejected; nothing is truncated for you.

For RAG or document indexing, chunk your text to well under 40,960 tokens per item before sending. Chunks of 512 to 1,024 tokens are typical for retrieval.

Every item in a batch must be non-empty. A blank or whitespace-only string fails the whole request. Filter empty strings before sending.

The dimensions parameter is not supported

OpenAI's text-embedding-3 models accept a dimensions parameter to shorten the output vector. Some client libraries and tools (for example Cherry Studio, and some Open WebUI setups) send it automatically.

qwen/qwen3-embedding-8b does not accept dimensions. Sending it fails the request. If your tool sets dimensions automatically, turn that option off. The model always returns the full 4096-dimensional vector.

Need smaller vectors? Qwen3-Embedding uses Matryoshka Representation Learning, so you can shorten the vector yourself: request the full 4096-dim embedding, keep the first N values, then re-normalise (L2). Normalising after truncation is what preserves quality.

import numpy as np

full = np.array(resp.data[0].embedding)   # 4096 dims
short = full[:512]                         # keep the first 512
short = short / np.linalg.norm(short)      # re-normalise

Errors

All of the cases below return the same generic response:

HTTP 400
{ "error": { "message": "The request contained an invalid argument.
  Please check your request parameters and try again.",
  "type": "invalid_request_error", "code": "400" } }

The API does not tell you which of these caused it, so check them in order:

Cause	What to change
Sent a dimensions parameter	Remove it; truncate client-side (see above)
Input over 40,960 tokens	Chunk the input to smaller pieces
Blank or whitespace-only item in a batch	Filter empty inputs before sending

See Also

• Models - full model catalogue
• Rate Limits
• API Examples