7 best mobile AI image generation apps for iPhone and Android in 2026

Key takeaways

• Gemini and ChatGPT both make images for free in their phone apps. Gemini runs Nano Banana 2, and ChatGPT moved every plan to Images 2.5 in September 2026.

• Adobe Firefly and Leonardo put several companies' models in one app, and both charge credits or tokens for the heavier ones.

• Free images on Leonardo and Ideogram are public. Pay, or pick another app, if your work needs to stay private.

• Draw Things is the on-device option: free, offline, and Apple-only, with larger models like FLUX needing a device with 8GB of memory.

• Ideogram and Draw Things have no Android app, and Midjourney has no official app on either platform, though its website works in a phone browser.

What is the best AI image generator app for a phone?

For most people, the Google Gemini app is the easiest place to start, because Nano Banana 2 is free in it on iPhone and Android. ChatGPT is the closest alternative and handles edits through conversation, while Draw Things is the one to use if you want images generated on the phone itself, with no connection.

Image apps on phones differ from each other in ways that are easy to miss. Some don't exist on Android. Some cap free use without saying by how much. A few make your free images public, and most mark their output, visibly or in metadata, in ways you'll want to know about before you post.

We've ranked these seven on whether they run natively on both iPhone and Android, how much the free tier allows, which models they offer, whether anything runs on the device, and what they attach to your images (watermarks, metadata, and public galleries).

The seven apps at a glance

1. Google Gemini: best free all-rounder

Google's Gemini app generates and edits images with Nano Banana 2, which needs no subscription.

Strengths: Free users get Nano Banana 2, and any Google AI plan adds Nano Banana 2.1, which Google ranks above it. The cheapest plan, Google AI Plus, costs $4.99 a month and doubles the standard limits; Pro at $19.99 quadruples them. The app's listing covers blending several photos into a mockup and posters with sharp text, plus resizing the result afterward.

Limits: Google no longer publishes a daily image count for any plan, only multiples of a "standard" limit. A visible watermark goes on images unless you switch off Media Watermark in settings, and the invisible SynthID mark stays either way. In India, South Korea, and Vietnam, only AI Ultra subscribers can turn the visible mark off.

Best for: Free, everyday image making and photo edits on either platform.

2. ChatGPT: best for editing through conversation

ChatGPT makes images inside a chat, so you refine a picture by describing what to change.

Strengths: OpenAI's ChatGPT Images 2.5, released in September 2026, runs on every plan, including free, on desktop, mobile, and web. On a phone you can tap an image, choose Select, and paint over just the part you want changed, and the mobile app lets you sketch a rough layout to guide a prompt. Go costs $8 a month and Plus $20.

Limits: OpenAI doesn't publish free image limits; the app shows them, along with reset times, when you get close. Free and Go accounts may see ads, and Plus and higher plans don't. OpenAI's help center warns that selected-area edits can spill past the area you marked. Images carry C2PA metadata and an invisible watermark.

Best for: Quick edits and variations you'd rather talk through than set up.

3. Adobe Firefly: best for several models in one app

Firefly is Adobe's standalone generator app, built for iPhone and also on Android.

Strengths: The app runs Adobe's own Firefly image models alongside partner models, and its App Store listing names Google, OpenAI, Luma, and Runway. Adobe's plans page starts Firefly Standard at $9.99 a month with 2,000 credits, and every paid plan includes unlimited standard image generations. Adobe attaches Content Credentials to output automatically.

Limits: Partner models cost credits per image. Adobe's current rate card lists Nano Banana 2 at 20 credits and Nano Banana Pro at 40, and marks most of those rates as limited-time. The free plan gives "limited daily generations" without a number, and you have to sign in to generate, with Google, Apple, Adobe ID, Microsoft, or LINE.

Best for: Trying Google's, OpenAI's, and Adobe's models without juggling separate apps.

4. Leonardo: best for ready-made templates

Leonardo started as a web suite and now has native apps on iPhone, iPad, and Android under one subscription.

Strengths: The iOS listing names Nano Banana, Seedream, FLUX, and Leonardo's own models, and the Android listing adds OpenAI's GPT image models. Blueprint templates turn a single tap into a finished style, which suits a small screen. The free plan refreshes 150 fast tokens a day, and the app can also animate generated images and photos into video.

Limits: Leonardo's pricing page lists the free plan's model access as "Auto Only," so you can't pick a model until you pay, and free creations are public. Essential is listed at $12 a month on the pricing page, while the iPhone app sells it at $15.99 a month.

Best for: Fast, styled results from presets, with a web account you can pick up later on a laptop.

5. Ideogram: best for text inside images

Ideogram built its name on legible lettering, and its iPhone app carries that over to logos, posters, and covers.

Strengths: The app generates, edits with generative fill, removes backgrounds, and makes variations of a character from a single uploaded photo. On the web, Ideogram 4.5 takes one to five reference images. Free accounts that sign in with Google, Apple, or Microsoft get 30 credits each week, according to Ideogram's plan docs.

Limits: There's no official Android app, so Android users work in the mobile browser. Accounts created with an email and password get no free credits, and everything made on the free plan is public. Plus costs $20 a month on the web against $28.99 a month inside the iPhone app.

Best for: Images where the words have to be spelled right.

6. Draw Things: best for private, on-device generation

Draw Things runs open image models on the iPhone or iPad itself, with no internet connection or subscription needed for local work.

Strengths: Local generation is free, with no ads or token limits, and prompts and images stay on the device. The App Store listing names SDXL, FLUX.1, FLUX.2, Qwen Image, Z Image, Wan 2.2, and LTX-2.3, and the app supports LoRAs, ControlNet, inpainting, and on-device LoRA training. An optional Draw Things+ plan at $8.99 a month adds cloud compute with a priority queue.

Limits: It's Apple-only and needs iOS 15.4 or later. The Draw Things wiki says 4GB of memory handles some smaller models but "Flux needs 8GB," and suggests SDXL Hyper at 8-bit for lower-memory devices. Models are big downloads: the wiki lists Qwen Image at 16GB in its 6-bit form. Its wiki says one image "might take seconds or minutes," depending on the device and model.

Best for: Private, offline generation and full control over open models.

7. Meta AI: best for quick images to share socially

The Meta AI app makes and edits images with Muse Image, the model Meta launched in July 2026.

Strengths: Meta says Muse Image "is free for everyday creation" in the Meta AI app and on meta.ai, and it also runs in Instagram Stories in the US. The app has style presets, and you can edit a photo by sketching on it or asking for a change. It's on both iPhone and Android.

Limits: You need a Meta account to log in, and Meta says the app isn't available everywhere yet without listing countries. No usage caps are published, and heavier use falls under paid plans. Images carry Meta's invisible Content Seal watermark. The iPhone app needs iOS 18.4 or later, the highest requirement of the seven.

Best for: Casual images headed for Instagram, Facebook, or WhatsApp.

Phone tips: storage, privacy, and exporting

• Check your iOS version first: Gemini needs iOS 18.2, ChatGPT and Firefly need iOS 18, and Meta AI needs 18.4, while Draw Things runs back to iOS 15.4.

• Budget storage for local models: The cloud apps take about 40MB to 390MB each on iPhone. Draw Things itself is 342MB, but every model it runs is a separate download, and the wiki lists Qwen Image at 16GB, so clear space before you start. Our guide to how much storage you need for local AI models has file sizes for the same model families.

• Know what's public: Free images on Leonardo and Ideogram appear publicly. Draw Things keeps local work on the phone, and the chat apps keep images in your account's history.

• Expect marks on exports: Gemini adds a visible watermark unless you turn it off, and SynthID regardless. ChatGPT adds C2PA metadata, Firefly adds Content Credentials, and Meta AI adds Content Seal.

• Subscribe on the web when you can: Leonardo's and Ideogram's in-app prices on iPhone run higher than the same plans bought in a browser.

• Keep characters consistent with references: Upload the same reference photo each time, since phone prompts tend to be short. Our ranking of AI image generators for keeping the same face covers which tools do this well.

Mobile AI image apps FAQ

What is the best free AI image app for iPhone and Android?

Google Gemini is the easiest free starting point for most people, since Nano Banana 2 costs nothing in the app and runs on both platforms. ChatGPT is close behind with Images 2.5 on its free plan, though OpenAI doesn't publish how many images that covers.

Can I generate AI images offline on my phone?

Yes, on an iPhone or iPad with Draw Things, which runs models on the device with no internet connection. There's no Android version. Larger models like FLUX need a device with 8GB of memory.

Is there an official Midjourney app?

No. Midjourney has no iOS or Android app, but midjourney.com works in a phone browser. Its docs say there's no free trial on the website, and Basic costs $10 a month. A limited trial is available in niji·journey, a separate iOS and Android app.

Do mobile AI image apps add watermarks?

Most add something. Gemini adds a visible watermark unless you switch it off in settings, plus an invisible SynthID mark that stays. ChatGPT, Firefly, and Meta AI use invisible marks or metadata, which don't change how the image looks.

Are my AI images private on these apps?

It depends on the plan. Free images on Leonardo and Ideogram are public, and paid plans make them private. Draw Things keeps local generations on your device.