QQwen Image 2.1Try Free

Home / How to Use

How to Use Qwen Image 2.1

Practical paths to run qwen image 2.1 — from a free browser demo to local ComfyUI and cloud APIs. No API keys are collected on this site.

Step 1 — Try free online (recommended first)

  1. Open the primary free demo: Hugging Face Space — Qwen/Qwen-Image-2.1
  2. Paste a prompt from our prompt gallery
  3. Generate, then iterate (add RGBA / edit / reference instructions as needed)
  4. Optional: try Qwen Cloud try-ai (qwen-image-2.1-pro)

Step 2 — Get official weights

Day-0 ecosystem support includes Diffusers, ComfyUI, vLLM-Omni, and SGLang.

Step 3 — Run with Diffusers (outline)

  1. Create a Python environment with a CUDA-capable PyTorch build (or CPU for tiny tests)
  2. Install the Diffusers stack versions recommended in the model card / GitHub README
  3. Load Qwen/Qwen-Image-2.1 and run a text-to-image sample, then an edit/RGBA sample
  4. For speed experiments, follow Turbo (~8 denoising steps) settings when documented

Always prefer the exact install commands from the model card — versions move quickly.

Step 4 — Run with ComfyUI

  1. Install / update ComfyUI
  2. Install the Day-0 Qwen Image 2.1 nodes/custom pack referenced by the official repo
  3. Download weights into the paths your workflow expects
  4. Start with a simple T2I graph, then add reference images (up to 10) or circle/annotation/mask local edits

Step 5 — Cloud API (Pro / Turbo)

Alibaba Cloud Model Studio published Pro + Turbo APIs around 2026-10-09. Use the official reference:

Qwen image generation and editing API reference

  • Create a Model Studio account and obtain credentials in the Alibaba console
  • Call generation / editing endpoints per the docs (Turbo for latency, Pro as offered)
  • Never paste secret keys into public Spaces or this website

Feature-focused tips

  • RGBA — explicitly ask for transparent background / RGBA for stickers and packshots
  • Local edits — mark regions with circles, annotations, or masks and describe only what should change
  • Identity — supply portrait references and state "preserve identity"
  • Typography — put the exact string in quotes inside the prompt

Read next

Prompt gallery · vs Flux-style models · Official Alibaba blog