On this page
You can run Qwen3-VL on your Android phone with Off Grid AI Mobile (OGAM). Search for Qwen3-VL, download Qwen’s own 2B or 4B build, and your phone reads bills, forms and screenshots with no internet and no uploads.
It’s free, there’s no account, and the 2B runs well on a 6 GB phone.
Google Play · Download Off Grid AI
What Qwen3-VL is
Qwen3-VL is Alibaba’s vision-language family, released in October 2025 under the Apache 2.0 licence. Qwen publishes its own GGUF files, the format OGAM runs, in 2B, 4B and 8B.
From Qwen’s model card, the parts that matter on a phone:
- Reads text in 32 languages, and copes with low light, blur and tilted pages.
- Understands app screens, so it can tell you which button does what.
- Finds objects in a picture and says where they are.
Which size fits your phone
OGAM judges fit from your phone’s RAM. On Android, a download under 52% of RAM runs well, 52% to 62% is marked Tight, and anything bigger isn’t offered. The download is the model file plus its vision file, because the app fetches both.
| Model | Total download (Q4_K_M + vision file) | 6 GB | 8 GB | 12 GB | 16 GB |
|---|---|---|---|---|---|
| Qwen3-VL 2B | 1.9 GB (1.11 + 0.82) | Runs well | Runs well | Runs well | Runs well |
| Qwen3-VL 4B | 3.3 GB (2.50 + 0.84) | Tight | Runs well | Runs well | Runs well |
| Qwen3-VL 8B | 6.2 GB (5.03 + 1.16) | No | No | Tight | Runs well |
Sizes come from Qwen’s GGUF repos. Android reports a bit less RAM than the box says, so a 12 GB phone lands the 8B in Tight rather than Runs well. To check your RAM, open Android Settings, then About phone.
Set it up
- Install OGAM from Google Play.
- Tap Models in the bottom bar. You land on the Text tab.
- Type Qwen3-VL in Search models.
- Tap Qwen3-VL-4B-Instruct-GGUF, or the 2B or 8B. Qwen’s own upload leads each card, and copies from other uploaders are counted in a +N uploads chip.
- Under Sizes, take the Recommended file and download it. Use Wi-Fi, because the 4B is about 3.3 GB.
- When it finishes, tap Chat now.
The cards say Newer: Qwen 3.5. OGAM is pointing you at the model it recommends today. Qwen3-VL still downloads and runs.
Once it’s downloaded, turn on airplane mode and send another message. It still answers, because the model runs on your phone.
Ask it about a photo or screenshot
Tap + in the message box, choose Photo, then Camera or Photo Library. Add your question and send.
- An electricity bill: “What’s the amount due, and what’s the due date?”
- A form in German: “Translate the field names and tell me which ones are required.”
- A screenshot of your bank app: “Which of these payments are subscriptions?”
- A shelf photo: “Is there any oat milk on this shelf, and where?”
A bank screenshot or an ID card is exactly what you don’t want on someone else’s server. Here, the photo goes from your gallery to the model in your phone’s memory and back to your screen.
Qwen3-VL or Qwen 3.5?
Qwen 3.5 is newer, reads images too and is what OGAM lists first. Pick Qwen3-VL when you mostly photograph text in other languages, or when a 4B vision model fits your phone better than Qwen 3.5’s 2B or 9B. For the app’s own pick, see the Qwen 3.5 Android guide.
If something goes wrong
| Symptom | Check | Fix |
|---|---|---|
| Qwen3-VL 8B is missing, or has no size to download | Your phone has under 12 GB of RAM | Pick the 4B |
| The app closes or the model won’t load | A Tight size, other apps holding memory | Close other apps, or pick the next size down |
| Photo says “Vision Not Supported” | The active model reads text only | Switch to the Qwen3-VL model you downloaded |
| The app says “Vision File Missing” | The model is there, its vision file isn’t | Open your downloads from that message, tap the wrench next to the model, then reload it |
| Small print comes back wrong | The photo is wide or blurry | Crop to the part you need and retake in good light |
FAQ
Can Qwen3-VL run on an Android phone? Yes. Search for Qwen3-VL in OGAM and download Qwen’s own GGUF upload. The app adds the vision file, and the model reads photos on the phone with no internet.
Which Qwen3-VL size fits my phone? The 2B on a 6 GB phone, the 4B on an 8 GB phone, and the 8B on a 16 GB phone. On a 12 GB phone the 8B loads but is marked Tight, so the 4B is the steadier pick.
How big is the Qwen3-VL download? With its vision file, the 2B is about 1.9 GB, the 4B about 3.3 GB and the 8B about 6.2 GB at Q4_K_M, the 4-bit build the app recommends.
Why does the app say Newer: Qwen 3.5 on Qwen3-VL? OGAM now recommends Qwen 3.5, which also reads images. The label is a heads-up, not a block. Qwen3-VL still downloads and runs.
Is it free and private? Yes. OGAM is free with no account. After the download, the model runs in your phone’s memory, so your photos aren’t uploaded to any server.
Related guides
- How to Run Vision AI Locally on Your Android Phone in 2026 (No Cloud, No Subscription)
- How to Ask AI About a Screenshot on Android in 2026 Without Internet
- How to Run LFM3 on Your Android Phone in 2026 (1.7B Chat and 2B Vision, Offline)
- How to Find AI Models Your Android Phone Can Actually Run in 2026 (No Failed Downloads)
Get Off Grid AI on Google Play and point Qwen3-VL at your next bill.