A direct line to your model.
Ask a question, see the answer arrive token by token, and keep the full OpenAI-compatible request within reach when you want to go deeper.
Instance
€0.80 / hour
Cost / 1M
~€0.11
Aggregate
2,000 tok/s
Per user
>95 tok/s
Concurrency
+20 users
Output / hour
7.2M tok
Estimated equivalent at full aggregate utilization for 20 concurrent users. QDivZero bills the instance by the hour, not by token.
Open the demo
Separate, time-limited test experience. Availability is subject to change.
QDivZero chat
nvidia/qwen3.6-35b-a3b-nvfp4-for-demo
Request JSONmodel, messages, stream and moreEdit
VS Code / Copilot
Bring QDivZero into Copilot.
Add the endpoint as a custom model in VS Code and use QDivZero from the Chat view. No API key is required for this demo.
- 01
Open Manage Language Models
In VS Code, open the Chat model picker and choose Add Models, then Custom Endpoint.
- 02
Choose Chat Completions
Use
https://qdiv0.com/api/qdiv0/v1and leave the API key asnot-needed. - 03
Paste the model definition
Save the generated
chatLanguageModels.jsonand select the model in Chat.
[
{
"name": "QDivZero",
"vendor": "customendpoint",
"apiKey": "not-needed",
"apiType": "chat-completions",
"models": [
{
"id": "nvidia/qwen3.6-35b-a3b-nvfp4-for-demo",
"name": "Qwen 3.6 35B A3B NVFP4 for demo",
"url": "https://qdiv0.com/api/qdiv0/v1",
"toolCalling": true,
"thinking": true,
"streaming": true
}
]
}
]Keep the API key in VS Code's secret storage. Tool calling requires a model and deployment that expose tool support.