HomeTemplates › Qwen3-235B-A22B
🧠 Self-hosted LLM

Qwen3-235B on a GPU block

Qwen's flagship MoE for reasoning and code

From
loading…
cheapest machine that meets this template
Deploy in one click →
Video memory
480 GB
Setup
~15 min
Access
port 8000
Billing
hourly, in BRL

Qwen3-235B is a mixture-of-experts model: many parameters in total, few active per token. That gives large-model quality with better throughput than the size suggests — as long as it fits in memory.

Qwen3-235B-A22B (MoE) served via vLLM with an OpenAI-compatible API — top open-source for reasoning and code. NOTE: needs multiple H100/H200 GPUs or quantization. For a single GPU, use the Qwen3 (8B) template.

What it is for

How to deploy

  1. Create your account and add balance (card or Pix, no subscription).
  2. In the console, pick the Qwen3-235B-A22B template and a machine — the console hides the ones that do not meet the requirement.
  3. In about 15 minutes the setup finishes and the access address shows up in the panel, on port 8000.

Done? Just destroy the machine and billing stops with it. No contract, no minimum commitment.

FAQ

How much video memory?

Around 480 GB — in practice, a block of 80–141 GB cards.

Does mixture-of-experts help the cost?

Yes: since only a few experts fire per token, per-request speed beats a dense model of the same size.

Can I run it quantised on fewer cards?

You can, at some quality cost. Use the console to see which blocks are available right now.

Run Qwen3-235B-A22B today
You only pay for the hours the machine is running.
Deploy in one click →

Related templates

vLLM
OpenAI-compatible API for Llama, Qwen, Mistral and more
TGI (HuggingFace)
HuggingFace Text Generation Inference — vLLM alternative
LiteLLM Proxy
1 OpenAI endpoint routing to 100+ providers (cloud + local)
Ollama
Run DeepSeek, Qwen3, Llama and Mistral with one command