Ann NgAnn Ng
5 min read

Use Qwen 3 Embedding 4B on Hugging Face with your own API key on TypingMind

Hugging Face Inference is a managed service that gives you access to thousands of open-source AI models — including Meta Llama, Mistral, Qwen, DeepSeek, and more — through a single OpenAI-compatible API.

Here is how to use Qwen 3 Embedding 4B (Qwen/Qwen3-Embedding-4B) on Hugging Face with TypingMind.

Qwen 3 Embedding 4B Overview

Model NameQwen 3 Embedding 4B
ProviderHugging Face
Model IDQwen/Qwen3-Embedding-4B
Release DateJan 1, 2025
Last UpdatedJan 1, 2025
Knowledge Cutoff2024-12
Context Window32,000 tokens
Max Output2,048 tokens
Pricing /1M tokens$0.01 input
$N/A output
Input Modalitiestext
Output Modalitiestext
Capabilities
Open Weights

Complete Setup Guide

1

Create a Hugging Face Account

Go to huggingface.co and create a free account if you don't have one. Hugging Face provides access to thousands of open-source AI models through their Inference API — no complex setup required.

2

Generate an Access Token

Go to your Hugging Face Access Tokens settings and click "+ Create new token".

Select Fine-grained as the token type and give it a name (e.g. "typingmind"). Under the Inference section, enable the following permissions:

  • ✅ Make calls to Inference Providers — required to use Hugging Face hosted models
  • ☑️ Make calls to your Inference Endpoints — optional, for dedicated endpoints
  • ☑️ Manage your Inference Endpoints — optional, for endpoint management
Create Hugging Face access token with Inference permissions

Click "Create token" and copy the generated token — you'll need it in the next step.

⚠️ Important: Keep your access token secure and never share it publicly. Store it safely as you'll need it to connect with TypingMind.

3

Add a Custom Model to TypingMind

Go to typingmind.com, click the Models tab on the left workspace bar, then click "Add Custom Model" in the top right corner.

TypingMind Models page with Add custom model button

Select "Create Manually" as the creation method.

Add Custom Model form in TypingMind

Fill in the following fields:

FieldValue
NameHuggingFace Qwen 3 Embedding 4B(or any name you prefer)
API TypeOpenAI Chat Completions API
Endpoint URLhttps://router.huggingface.co/v1/chat/completions
Model IDQwen/Qwen3-Embedding-4B
Context Length32,000for Qwen 3 Embedding 4B

Scroll down to the Authentication section and set:

FieldValue
Authentication TypeAPI Key via HTTP Header
Header KeyAuthorization
Header ValueBearer {{API key}}
Authentication settings in TypingMind

Under Model Capabilities, enable the features supported by your chosen model:

  • ✅ Support System Role — recommended for most models
  • ✅ Support Streaming Output — recommended for all models
  • ☑️ Support Plugins — enable if the model supports function/tool calling
  • ☑️ Support Image Input — enable for vision-capable models
  • ☑️ Support Thinking — enable for reasoning/thinking models
Model Capabilities settings in TypingMind

After filling in all fields, click "Test". If everything is set up correctly, you will see the message "Nice, the endpoint is working!"

Click "+ Add Model" to finish.

4

Use the Model

Now you can select your newly added Hugging Face model from the model picker and start chatting!

  • Select Qwen 3 Embedding 4B from the model dropdown menu
  • Start typing your message in the chat input
  • Enjoy faster responses and better features than the official interface
  • Switch between different AI models as needed
typingmind.com
Chat with Qwen 3 Embedding 4B via Hugging Face on TypingMindHugging FaceHugging FaceHuggingFace Qwen 3 Embedding 4B

💡 Pro tips for better results:

Hugging Face FAQ

Why use Qwen 3 Embedding 4B on Hugging Face with TypingMind instead of the official Hugging Face interface?

Hugging Face's web interface is built for model exploration and research — not for everyday AI chat. TypingMind gives you a polished, full-featured chat interface on top of Hugging Face Inference: conversation history, prompt library, AI agents, plugin support, file uploads, and more. You bring your own HF access token and pay only for what you use, with no TypingMind subscription required.

Do I need a Hugging Face subscription to use Qwen 3 Embedding 4B?

No subscription required for many models. Hugging Face offers a free tier with rate limits, and a PRO plan ($9/month) for higher usage. Many open-weight models are completely free to use. You only pay for inference on paid models. TypingMind is also a one-time purchase, not a subscription.

How much does it cost to use Qwen 3 Embedding 4B via Hugging Face?

Qwen 3 Embedding 4B on Hugging Face costs $0.01/1M input tokens and $0/1M output tokens. Many open-weight models are free to use. You can monitor usage in your Hugging Face billing dashboard.

What is Hugging Face Inference and how does it work with TypingMind?

Hugging Face Inference is a managed service that lets you run thousands of open-source AI models via a simple OpenAI-compatible API. TypingMind lets you connect to Hugging Face using your own access token, so you can chat with models like Qwen 3 Embedding 4B directly — without any intermediary subscription.

Can I use other Hugging Face models besides Qwen 3 Embedding 4B in TypingMind?

Yes! Hugging Face gives you access to thousands of open-source models from Meta, Mistral, Qwen, DeepSeek, and more. You can add multiple models as custom models in TypingMind and switch between them instantly in any conversation.

Is my data private when using Hugging Face with TypingMind?

Yes. TypingMind stores your conversations locally — in your browser (web) or on your device (desktop app). Hugging Face processes inference requests through their secure infrastructure. Your data never passes through TypingMind servers.

What token permissions do I need for the Hugging Face access token?

You need a Fine-grained token with "Make calls to Inference Providers" permission enabled under the Inference section. Optionally, you can also enable "Make calls to your Inference Endpoints" if you use dedicated endpoints.

Can I use Qwen 3 Embedding 4B on Hugging Face for commercial projects?

It depends on the model's license. Qwen 3 Embedding 4B is an open-weight model — check its specific license on Hugging Face for commercial use terms. TypingMind supports commercial use.