Home / Compare / Best AI image generators
The best free AI image generators, and what the paid ones cost
Three of the biggest AI apps make pictures for free: Google Gemini, ChatGPT and Microsoft’s Bing Image Creator. Bing is the most open about its limits, and an open model you run on your own computer has none at all. Here is what each free option gives you, when paying is worth it and what images cost through an API.
Checked October 18, 2026 against each company’s own plans, help and pricing pages
Quick picks
Free AI image generators compared
| Free model | Published free limit | Edits your photos | Paid plans, a month | |
|---|---|---|---|---|
| Bing Image Creator | Microsoft’s MAI-Image-2.5-Flash and MAI-Image-2e | 15 fast creations a day, then slower ones; up to 200 prompts a day | Yes | None; Microsoft Rewards points buy more fast creations |
| Google Gemini | Nano Banana 2 | “Standard limits”, not given as a number; paid plans get 2 to 20 times more | Yes, and combines two photos into one | Google AI Plus $4.99, Google AI Pro $19.99, Google AI Ultra from $99.99 |
| ChatGPT | OpenAI’s image generator | “Limited”, not given as a number; more on Go and Plus, unlimited on Pro | Yes | Go $8, Plus $20, Pro from $100 |
| Grok | Grok Imagine | “Limited”, not given as a number | Not listed on its plans page | Plus $8, Premium $30 (adds video), Super $100, Ultra $200; the plans also cover X and Cursor |
| Ideogram | Ideogram’s own image models | 30 credits a week if you sign in with Google, Apple or Microsoft; free images are public | Yes | Plus $20 ($15 billed yearly), Pro $60 ($42), Team $30 a user ($20) |
| Canva | AI image tools inside its design editor | Up to 20 Standard or Premium AI uses | Background removal and other AI edits on Pro | Pro $144 a year; Business $250 a year a person |
| Leonardo | Its own Phoenix and Lucid models, among others | 150 fast tokens a day; free creations are public | Not listed on its plans page | Essential $12, Premium $30, Ultimate $60; up to 20% off yearly |
| FLUX 3 Image Klein | Open weights you download | None: it runs on your own computer | Yes | None; the same model costs $0.009 an image through the API |
You need a free account for each app: a Microsoft account for Bing (work and school accounts aren’t accepted), a Google account for Gemini, an OpenAI account for ChatGPT.
Bing Image Creator publishes the clearest rules: 15 fast creations a day that come back the next day, then unlimited slower ones, up to 200 prompts in 24 hours. Each creation can give you several pictures. You can upload a photo to edit it, and switch on a visible watermark that marks the result as AI-made.
Google Gemini makes and edits images with Nano Banana 2 on the free plan. Google says Gemini can combine two photos into one, and that its Nano Banana models put clear text into pictures in many languages, for invitations, posters and logos.
ChatGPT makes pictures from a description or changes a photo you upload on the free plan. OpenAI calls free image creation “limited” without giving a number.
What paying adds
Gemini: any paid Google AI plan unlocks Nano Banana 2.1, Google’s newest image model, and lets you redo a picture with Nano Banana Pro. It also adds video generation. Google AI Plus is the cheapest at $4.99 a month, with twice the free usage; Google AI Pro, at $19.99, has four times.
ChatGPT: Go ($8 a month) gives more image creation than Free; Plus ($20) adds “more complex and accurate image creation” and image generation with Thinking; Pro (from $100) gives unlimited and faster image creation, OpenAI’s pricing page says. See also Gemini vs ChatGPT.
Bing: there’s nothing to buy. Microsoft Rewards points, earned by searching with Bing, pay for extra fast creations.
Ideogram: Plus costs $20 a month ($15 billed yearly) for 2,400 credits a month, unlimited slow generations, private images and quality export; Pro, at $60 ($42), has 7,500 credits and batch generation. How many credits an image takes depends on the model, quality and size. Extra credits cost 1 cent each.
Grok: image generation with Grok Imagine comes with every paid plan, starting with Plus at $8 a month, which shares one AI usage pool across Grok’s chat, voice, Imagine and Build. Video generation starts at Premium ($30), which has four times Plus’s usage. The same subscription also covers X and the Cursor code editor.
Adobe Firefly: Standard costs $9.99 a month for 2,000 generative credits and unlimited standard image features such as Generative Fill. Pro, at $19.99, has 4,000 credits, Adobe Express Premium and Photoshop on web and mobile, and for the first year unlimited generations on select models, including Google’s Nano Banana 2 at 1K and FLUX.2 Pro. Pro Plus ($49.99) makes every image model and some video models unlimited at 1K and 2K for the first year, and Premium ($199.99) extends that to 4K and more video models. Firefly also runs models from Google, OpenAI and others.
Canva: Pro, at $144 a year, gives 10 times the free plan’s AI uses, and Business, at $250 a year per person, 20 times; an AI Pass add-on costs extra.
Leonardo: Essential costs $12 a month for 8,500 fast tokens; Premium, at $30, has 25,000 plus unlimited image generation at a relaxed pace on selected models; Ultimate, at $60, has 60,000 and adds relaxed video. Paying yearly saves up to 20%, and team plans start at $24 a seat.
If you only make a few pictures for cards or social posts, the free plans are usually enough. Paying makes sense when you hit the limits often or want Google’s newer models.
Midjourney: no free plan, unlimited images from $30
Midjourney has no free plan: all four of its plans are subscriptions. Every plan includes some fast generation time; from Standard up you can also make unlimited images in Relax mode, which is slower. Yearly billing saves 20%.
| Plan | Monthly | Billed yearly | Fast time a month | Relax mode |
|---|---|---|---|---|
| Basic | $10 | $8 a month | About 3.3 hours (200 minutes) | None |
| Standard | $30 | $24 a month | 15 hours | Unlimited images |
| Pro | $60 | $48 a month | 30 hours | Unlimited images and SD video |
| Mega | $120 | $96 a month | 60 hours | Unlimited images and SD video |
Extra fast time costs $4 an hour. Stealth mode, which keeps your images and videos private, comes only with Pro and Mega. All plans allow commercial use, but a company with more than $1 million a year in revenue must buy Pro or Mega.
Free with no limit: an image model on your own computer
Black Forest Labs released FLUX 3 Image Klein on October 15, 2026 as open weights: a 5.6-billion-parameter model you can download and run yourself, made to work on consumer graphics cards. It makes and edits images and can take up to 10 reference pictures. Its license allows free commercial use for individuals and companies with less than $5 million in yearly revenue.
You need a computer with a capable graphics card and an app such as ComfyUI to run it. Without that, the same model costs $0.009 an image through Black Forest Labs’ API.
AI image generation prices through an API
Developers and anyone making pictures in bulk pay per image. These are the makers’ list prices:
| Model | Maker | Cheapest setting | 100 images | Bigger or better |
|---|---|---|---|---|
| GPT Image 2 | OpenAI | $0.006low quality, 1024 × 1024 | $0.60 | $0.053 medium, $0.211 high |
| FLUX 3 Image Klein | Black Forest Labs | $0.0091K | $0.90 | $0.014 at 2K, $0.036 at 4K |
| Grok Imagine Image | xAI | $0.02one size | $2.00 | $0.04 for Grok Imagine Image 2.0 |
| Nano Banana 2.1 | $0.03361K | $3.36 | $0.050 at 2K, $0.113 at 4K | |
| FLUX 3 Image | Black Forest Labs | $0.0481K | $4.80 | $0.10 at 2K, $0.607 at 4K |
| Nano Banana Pro | $0.1341K or 2K | $13.40 | $0.24 at 4K |
Prices are for the finished picture; the prompt and any photo you send in cost a little extra. Low-quality and full-quality settings aren’t comparable, so test a few prompts before choosing. OpenAI’s newer GPT Image 2.5 bills per token ($30 per million image output tokens), so its cost per picture depends on size and quality.
What happens to the photos you upload
On Ideogram’s and Leonardo’s free plans, what you make is public; their paid plans keep it private. Bing Image Creator says it doesn’t use uploaded photos to train models, deletes photos with faces within 30 days and keeps your creations for up to 90 days. Gemini: Google’s privacy hub says that if your Keep Activity setting is on, photos you share from your gallery are used to improve its services, with the help of human reviewers. ChatGPT: see OpenAI’s privacy policy. Upload photos of other people only with their consent.
What people say
Three recent launches drew long discussions on Hacker News: FLUX 3 Image, Krea 3 and FLUX 3 Image Klein.
- Control. Commenters liked tools that place each element exactly where you want it, rather than describing everything in a chat.
- Small text. A recurring complaint: edits still scramble small text and signs.
- Open weights. Many want models they can run themselves. FLUX 3 Image Klein got mixed reviews: good at editing, weaker on fine text, and some found it clearly behind the full FLUX 3 model.
- Editing photos of people. One tester asked Krea 3 to seat a woman in a T-shirt at a bar; the shirt’s print ran over an invented jacket, and they found Gemini did much better.
- Guardrails. Some said strict filters block harmless requests.
For rankings, LMArena and Artificial Analysis rate image models from blind comparisons of their pictures. Their order changes often, so check them when you choose.
Questions
What is the best free AI image generator?
For the most free pictures, Bing Image Creator: 15 fast creations a day, then slower ones up to 200 prompts a day. For editing your own photos, Gemini or ChatGPT. For no limit at all, run FLUX 3 Image Klein on your own computer.
Is there an AI image generator with no limits?
Not for free among the big online apps: each caps free use. Bing keeps making images at a slower speed after your 15 fast ones, up to 200 prompts a day. Adobe Firefly Pro ($19.99 a month) makes unlimited images on select models for its first year, and for a flat $30 a month Midjourney Standard and Leonardo Premium make unlimited images at a slower pace. An open model on your own computer has no cap beyond your hardware.
What is the cheapest AI image generator?
The free apps cost nothing. Through an API, OpenAI’s GPT Image 2 at low quality costs $0.006 a picture and FLUX 3 Image Klein $0.009 at 1K; Google’s Nano Banana 2.1 costs $0.034 at 1K.
Is Midjourney free?
No. Its plans start at $10 a month ($8 billed yearly) for about 200 minutes of fast generation. Standard, at $30, adds unlimited images in the slower Relax mode.
Can I use AI images for my business?
Each service sets its own terms, so read them before you sell or print anything. Midjourney allows commercial use on all plans but requires Pro or Mega for companies with more than $1 million a year in revenue; FLUX 3 Image Klein’s license allows free commercial use for individuals and companies under $5 million in yearly revenue.
Sources
- Bing Image Creator help (models, daily limits, uploads, watermark)
- Google: Gemini image generation and Gemini Apps limits and features (free model, paid models, usage)
- Gemini subscriptions, ChatGPT pricing and ChatGPT on the App Store (plans, free features)
- Midjourney: comparing plans, Ideogram: plans and credits, Leonardo pricing, Canva pricing, Adobe Firefly plans and Grok plans
- Black Forest Labs: FLUX 3 Image Klein and FLUX pricing
- OpenAI: image generation guide, Gemini API pricing and xAI models and pricing (prices per image)
- Gemini Apps Privacy Hub
- Hacker News launch discussions, linked above