What is Qwen AI?
Qwen AI is Alibaba Cloud's family of large language models and the consumer apps that let you chat with them. The model family includes Qwen3-Max (the flagship trillion-parameter model), Qwen3-Plus (the everyday workhorse), Qwen3-Coder for software engineering, Qwen-VL for vision, and Qwen-Omni for voice and audio. The consumer-facing app, branded as "Qwen Chat" or "Qwen Studio" depending on the version, brings the entire family into a single interface that runs on Windows, macOS, iOS, Android, and the web.
What makes Qwen download-worthy in 2026 is the combination of completely free access to frontier-class AI (no subscription, no usage caps that matter for most people) and genuinely good multimodal capabilities voice, vision, document analysis, image generation, video generation, and Deep Research are all bundled in the same free download. Most competitors gate at least half of these features behind paid plans starting at $20–200 per month. Qwen's strategy is to give the consumer experience away free and monetize through the developer API and enterprise contracts, which means you get the full app at no cost.
This article covers every legitimate Qwen AI download path: the official apps for each platform, the developer SDK and API setup, and the open-weight model downloads for users who want to self-host. If you want to get Qwen on whatever device you're using right now, the next section has the direct links.
Download Qwen AI for Your Platform
Pick the platform you're on. The same Qwen account works across every app, so once you sign in your chat history syncs automatically wherever you install.
Windows
Native desktop app with screen reading, file drag-and-drop, and multi-window support for professional workflows.
macOS
Universal binary, optimized for Apple Silicon. Native macOS integration with Spotlight and Shortcuts.
iOS / iPadOS
Native app for iPhone and iPad. Camera, voice chat, and document upload all work natively.
Android
Play Store install or APK sideload for unsupported regions. 5M+ downloads globally.
Web App
No download required. Works in Chrome, Safari, Firefox, Edge on any device. Same account as the native apps.
Developers
DashScope API for hosted access, plus open-weight model downloads from Hugging Face and ModelScope.
💡 The single official source for every download is qwen.ai/download. From there you'll find platform-specific install pages, system requirements, and release notes. For Android in regions without Play Store access, the trusted mirror at qwen.en.uptodown.com hosts the same official signed APK.
Download Qwen for Windows
The Windows desktop app is a full-featured native client built for professional and power users. Beyond what the web app offers, it adds screen reading (highlight content anywhere and ask Qwen about it), file drag-and-drop for instant uploads, multi-window chat sessions, and tighter integration with the Windows file system.
System requirements
- OS: Windows 10 (build 19041 or later) or Windows 11
- CPU: Any 64-bit x86 processor (Intel or AMD)
- RAM: 4 GB minimum, 8 GB recommended
- Storage: ~250 MB for install, plus ~1 GB free for chat history and cached media
- Network: Active internet connection (models run in Alibaba's cloud)
Install steps
- Go to qwen.ai/download and click the Windows download button.
- Save the
Qwen-Setup-x.x.x.exeinstaller to your Downloads folder. - Double-click the installer to launch it. If Windows SmartScreen warns about an unrecognized publisher, click More info → Run anyway (the installer is properly signed; SmartScreen is just cautious about new apps).
- Choose your install location and Start menu folder, then click Install.
- Launch Qwen from your Start menu or desktop shortcut. Sign in with Google, GitHub, or email all free.
- Pick your model from the selector in any new chat and start using it.
Download Qwen for macOS
The macOS app is a universal binary that runs natively on both Apple Silicon (M1/M2/M3/M4) and Intel Macs. Apple Silicon performance is noticeably better the app launches faster and feels snappier. Beyond chat, the Mac app integrates with macOS Shortcuts and Spotlight, so you can wire Qwen into broader workflows on the desktop.
System requirements
- OS: macOS 12 (Monterey) or later
- Architecture: Apple Silicon (recommended) or Intel x86_64
- RAM: 4 GB minimum, 8 GB recommended
- Storage: ~280 MB for install, plus ~1 GB working space
- Network: Active internet connection required
Install steps
- Visit qwen.ai/download and click the macOS download option.
- Open the downloaded
Qwen-x.x.x.dmgfile. - Drag the Qwen icon into your Applications folder.
- For the first launch, right-click the app in Applications and choose Open to bypass macOS Gatekeeper. Confirm the warning. Subsequent launches don't require this.
- Sign in with Google, GitHub, or email.
- Optionally, drag the Qwen icon to your Dock for quick access.
Download Qwen for iOS / iPadOS
The iOS app is published by Alibaba Group and available worldwide on the App Store (with rare regional exceptions). It's optimized for both iPhone and iPad, with full Pencil and external keyboard support on iPad.
System requirements
- OS: iOS 15+ / iPadOS 15+
- Devices: iPhone 8 or later, all iPad models from 2018 onward
- Storage: ~100 MB for install
Install steps
- Open the App Store on your iPhone or iPad.
- Search for "Qwen" and look for the app published by Alibaba Group. Verify the publisher carefully look-alike apps exist.
- Tap Get and authenticate with Face ID, Touch ID, or your Apple ID password.
- Launch from your home screen and sign in.
- Grant permissions for microphone (voice chat), camera (photo analysis), and notifications as you encounter the features that need them.
If the App Store doesn't show Qwen in your country, the web app at chat.qwen.ai works identically through Safari. Add it to your home screen via the share button → "Add to Home Screen" for an app-like experience.
Download Qwen for Android
The Android app has crossed 5 million downloads globally and is the highest-rated Qwen client at 4.07 stars. Two install paths depending on whether you have Play Store access.
System requirements
- OS: Android 8.0 (Oreo) or later
- Architecture: arm64-v8a (most modern phones), armeabi-v7a, x86_64
- Storage: ~120 MB for install
- RAM: 2 GB minimum, 4 GB recommended
Option 1 Google Play Store
- Open the Play Store on your device.
- Search for "Qwen" and find the app from Qwen Team / Alibaba Group.
- Tap Install.
- Launch the app and sign in.
Option 2 Sideload APK (unsupported regions)
- On your phone, open qwen.en.uptodown.com/android in your browser.
- Tap Download.
- Open Android Settings → Apps → Special access → Install unknown apps and allow your browser to install APKs temporarily.
- Open the downloaded
.apkfile from your Downloads folder. - Tap Install, then launch and sign in.
- Revoke the install permission afterward for security.
The Uptodown APK is the same official Alibaba-signed build as the Play Store version it's not a third-party fork. For full details on safe APK installation, including verification steps, see the comprehensive APK guide.
Use Qwen on the Web (No Download)
If you don't want to install anything, the web app at chat.qwen.ai works in any modern browser and includes essentially every feature the native apps offer chat, voice, vision (via webcam or upload), document analysis, image generation, video generation, and Deep Research. It's particularly useful for:
- Trying Qwen with zero commitment no install, no permissions, no disk space used.
- Public computers or work machines where you can't install software.
- Chrome OS, Linux, and other platforms that don't have a native app yet.
- Quick one-off questions when you don't want to launch a separate app.
The web app supports the same login methods (Google, GitHub, email) and your conversation history syncs with the native apps if you sign in on both. For long-term daily use, the native apps offer better integration and performance, but for everything else the web app is fully sufficient.
Download for Developers
Developers have two distinct download paths depending on whether they want to call Qwen over the API or run the models locally.
Path 1 API access (cloud-hosted)
The simplest developer path is to get an API key for Alibaba Cloud Model Studio (DashScope) and use Qwen as a hosted service. No model download required your code talks to Alibaba's servers.
- Sign up at alibabacloud.com (separate from the consumer Qwen Chat account).
- Activate Model Studio at alibabacloud.com/en/product/modelstudio. This enables the 1M+1M token free quota.
- Generate an API key from the console under API Keys. It starts with
sk-. - Install the OpenAI Python SDK:
pip install openai - Point your client at the DashScope endpoint and start making requests:
from openai import OpenAI
import os
client = OpenAI(
api_key=os.getenv("DASHSCOPE_API_KEY"),
base_url="https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
)
response = client.chat.completions.create(
model="qwen-plus",
messages=[{"role": "user", "content": "Hello, Qwen!"}]
)
print(response.choices[0].message.content)
Path 2 Open-weight model downloads (self-host)
The entire Qwen family ships with openly downloadable weights for self-hosting. This is the path for users who want full control, no API costs, and no data leaving their infrastructure. Models are hosted on:
- Hugging Face huggingface.co/Qwen primary distribution channel for the West. Every Qwen3 model variant lives here.
- ModelScope modelscope.cn/organization/qwen Alibaba's own model hub. Faster downloads from China.
- GitHub github.com/QwenLM inference code, training scripts, and documentation.
- Ollama
ollama run qwen3easiest way to run smaller Qwen variants locally on consumer hardware.
Available open-weight models include the full Qwen3 family (dense 0.6B / 1.7B / 4B / 8B / 14B / 32B and MoE 30B-A3B / 235B-A22B), Qwen3-Coder, Qwen-VL, and the older Qwen2.5 series. Most are released under Apache 2.0 license, which permits unrestricted commercial use.
Example using Hugging Face Transformers to load Qwen3-8B locally:
pip install transformers torch accelerate
from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "Qwen/Qwen3-8B"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype="auto",
device_map="auto",
)
prompt = "Explain transformers in one paragraph."
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
Hardware requirements for self-hosting
| Model | Minimum VRAM | Recommended Hardware |
|---|---|---|
| Qwen3-0.6B / 1.7B | 2-4 GB | Any modern laptop / phone |
| Qwen3-4B / 8B | 8-16 GB | RTX 3060 / Apple M2 Pro |
| Qwen3-14B | 24 GB | RTX 4090 / RTX 3090 |
| Qwen3-32B | 48-64 GB | 2× RTX 4090 or A6000 |
| Qwen3-30B-A3B (MoE) | 32 GB | RTX 4090 (only 3B active per token) |
| Qwen3-235B-A22B (MoE) | 140+ GB | 4× H100 or 8× A100 |
Verify Your Download Is Authentic
Before installing any Qwen download, a quick verification step protects against phishing and tampered builds.
For consumer apps
- Source check: Only download from qwen.ai/download, the official App Store / Play Store listings, or trusted mirrors (Uptodown, APKPure, Aptoide).
- Publisher name: "Alibaba Group" or "Qwen Team" anything else is suspicious.
- File size: Within 10–20% of the sizes listed above. Dramatically smaller or larger files are red flags.
- Avoid "Pro" / "Mod" / "Premium" variants: Qwen is already free; these are all fake and typically contain malware.
For developer downloads
- Hugging Face verified badge: The official Qwen organization on Hugging Face has the verified checkmark. Download only from
huggingface.co/Qwen, not from third-party mirrors. - SHA-256 checksums: Hugging Face displays file hashes; verify the downloaded file matches before loading it into your model server.
- GitHub repository activity: The official QwenLM organization on GitHub has frequent commits, official releases, and an Alibaba-verified account.
⚠️ Common scam: Sites offering "Qwen for Windows direct download" with prominent ads, surveys, or "click here to unlock" gates are not legitimate. The real Windows download comes from qwen.ai/download with no friction. If you're being asked to complete offers or install other software first, close the tab.
What You Get With Every Qwen Download
Regardless of which platform you download to, you get access to the same core feature set:
- Chat with Qwen3-Max and the entire Qwen3 model family via in-app picker.
- Voice conversations powered by Qwen3-Omni with real-time speech.
- Camera vision (mobile) and image upload (all platforms) for visual Q&A.
- Document analysis PDFs, Word docs, spreadsheets, CSVs.
- Image generation via Qwen-Image, including precise text rendering.
- Video generation via Wan 2.5+ text-to-video and image-to-video with native audio.
- Deep Research with cited reports, shareable webpages, and multi-host podcasts.
- Real-time web search with cited sources.
- Slide generation research-to-PowerPoint workflow.
- 119+ languages for chat, translation, and document processing.
- Cross-device sync start a chat on phone, continue on laptop.
- Free forever at the consumer tier no credit card, no subscription.
Common Download Issues
Download stops or times out
Slow connections sometimes can't complete the 100–300 MB download in one session. Most browsers and download managers support resuming pause and resume the download rather than restarting. For sideload APKs, downloading to your computer first and transferring to your phone via USB is often faster than direct mobile download.
"This file may be harmful" warning
Most browsers warn about installer downloads from sources they haven't seen before. This isn't an actual virus alert it's a precaution. For Qwen, the warning is safe to dismiss if you downloaded from qwen.ai/download or one of the trusted mirrors. Always check the URL before dismissing security warnings.
Installer won't run on Windows
Windows SmartScreen sometimes blocks new installers. Click More info → Run anyway to proceed. The installer is properly code-signed; SmartScreen is just cautious about apps it hasn't seen distributed widely yet.
"App is damaged and can't be opened" on Mac
macOS Gatekeeper occasionally flags downloaded apps. Right-click the app in Applications, choose Open, and confirm the dialog. After the first successful launch, normal double-click works going forward.
Play Store says "this item isn't available in your country"
Use the APK sideload path from Uptodown. The official Alibaba-signed APK works in essentially every region regardless of Play Store restrictions.
Hugging Face download is slow
For users in mainland China or with slow connections to Western servers, ModelScope (Alibaba's own model hub) mirrors all Qwen weights and is usually 5–10× faster. Switch your download command from huggingface-cli to ModelScope's equivalent and the gigabytes fly down.
FAQ
Is the Qwen AI download really free?
Yes. Every Qwen consumer app Windows, Mac, iOS, Android, web is completely free with no in-app purchases for core features. The open-weight model downloads are also free under the Apache 2.0 license. Paid options exist (the Coding Plan subscription, the developer API, enterprise contracts) but they're separate from the free downloads.
Where is the official Qwen download page?
qwen.ai/download is the single canonical download hub. It links to the Windows installer, Mac DMG, iOS App Store, Android Play Store, and the web app. Bookmark it.
What's the difference between Qwen Chat and Qwen Studio?
Two names for related apps. Qwen Chat is the older / international branding. Qwen Studio is the current package name (ai.qwenlm.chat.android on Android). Both are official Alibaba apps with the same underlying functionality. The web app and macOS/Windows versions are most commonly called "Qwen Chat," while the Android version increasingly shows up as "Qwen Studio."
Does Qwen work offline?
The official Qwen apps require an internet connection because models run in Alibaba's cloud. For genuine offline use, download the open-weight models from Hugging Face and run them with Ollama, llama.cpp, or vLLM on your own hardware. Qwen3-4B and smaller can run on consumer laptops; larger models need dedicated GPUs.
Will the Qwen app slow down my computer?
No. Because the AI models run in Alibaba's cloud, the local app is a lightweight chat client similar in resource usage to a Slack or Discord client. Idle CPU and memory use is minimal; active CPU only spikes briefly when sending and receiving messages.
How big are the model downloads?
The open-weight models range from about 1 GB (Qwen3-0.6B) to roughly 470 GB (Qwen3-235B in full precision). Quantized versions are much smaller Qwen3-8B in 4-bit quantization is around 5 GB. For most self-hosting use cases, the 8B and 14B variants in 4-bit or 8-bit are the sweet spot.
Can I install Qwen on Linux?
Alibaba doesn't currently publish an official Linux desktop app. Linux users have two great options: the web app at chat.qwen.ai works perfectly in Firefox, Chromium, or any modern Linux browser, or community projects let you run the official Windows version under Wine. For self-hosting, Linux is the preferred OS every inference framework (vLLM, Ollama, llama.cpp, Transformers) has first-class Linux support.
Is the Qwen download safe?
The official downloads from qwen.ai, the App Store, the Play Store, Hugging Face, and trusted mirrors are safe. They're code-signed by Alibaba and scanned by the relevant platform's security systems. Downloads from random third-party "Qwen Pro" sites are almost universally unsafe stick to the official sources.
Does the Qwen app collect my data?
The app sends your prompts and uploaded files to Alibaba Cloud for processing that's how the AI works. Per the Terms of Service, free-tier interactions may be used for service improvement. The app does not collect contacts, SMS, location, or other unrelated device data. You can review exact permissions in your OS settings and delete chat history from in-app settings.
How do I uninstall Qwen?
Windows: Settings → Apps → Installed apps → Qwen → Uninstall. macOS: drag Qwen from Applications to Trash. iOS: long-press the app icon → Remove App. Android: long-press → Uninstall. Your account and chat history are preserved on Alibaba's servers reinstall any time and sign in to restore everything.
Final Thoughts
Qwen AI's download experience is unusually clean for a frontier AI product in 2026 official apps for every major platform, an open-weights track for developers who want full control, and a web option for anyone who doesn't want to install anything. Best of all, every consumer download is free with no subscription wall, which makes Qwen the easiest serious AI to actually try.
Start at qwen.ai/download, pick the platform you're using, install, sign in, and you'll be chatting with Qwen3-Max in under five minutes. If you change your mind about which app you prefer, your account and chat history travel with you install on a different device, sign in, and pick up exactly where you left off.