Adobe is the undisputed king of photo and design software, and that alone makes its Firefly AI image and video generator worth your attention. The web app packages some nifty tools for creating art, faux photography, or short video clips to taste in a top-notch interface, and Adobe stresses that it doesn't use your input for training. The standalone Firefly experience also goes beyond the generative features it powers within Adobe's Express, Photoshop, and Illustrator. It's a good option if you're already deep in the Creative Cloud ecosystem, though we wish it could produce legible text and let you refine creations with follow-up requests. We've yet to determine an Editors' Choice winner in this rapidly evolving category, but third-party services like OpenAI Sora and Stable Diffusion remain ahead at this point.

How Do You Get Firefly?

You can try out Firefly in its purest form at firefly.adobe.com. There, you can see all the shiniest, newest capabilities. You can also access some Firefly features in Adobe Stock, Express, InDesign, Illustrator, and Photoshop, as mentioned. For example, in Photoshop, you can use Firefly to replace the background of an image with an entirely new one that you describe in words. This review focuses on the standalone version of Firefly on the web.

(Credit: Adobe/PCMag)

Spec-wise, you need ChomeOS, macOS 12 or later, or Windows 10 or later to run Firefly. Adobe doesn’t mention any CPU or graphics requirements, only that you must have at least 4GB of RAM. Supported browsers include current versions of Chrome, Edge, Firefox, and Safari. You can also run it on a mobile device running Android 9 or later and iOS 17.4 or later, but there are no mobile Firefly apps.

How Much Does Firefly Cost?

Since generative AI is such an intense computational activity that requires processing on Adobe’s servers, you don’t get unlimited access to it. Instead, a subscription to Firefly comes with Generative Credits. Most generating actions use one credit for standard images of up to 2,000 by 2,000 pixels. For comparison, Microsoft gives you Boosts for Copilot, which speeds up image generation; after you run out of Boosts, your images come slowly.

You can use the Firefly web app for free once you create an account or with a paid Creative Cloud subscription. Adobe isn't specific about how many credits free users get, stating that it's just enough to get a feel for the capabilities. Results get a watermark, but at least the interface doesn’t show ads.

Adobe offers two standalone subscriptions for Firefly: Firefly Standard costs $9.99 per month ($99.99 per year) and includes 2,000 generative credits per month, which is enough to create 20 five-second videos or translate six minutes of audio and video. A Firefly Pro subscription goes for $29.99 per month ($299.99 per year). It includes 7,000 credits per month, good for 70 videos or 23 minutes of translation. The company plans to offer a Premium level for corporate use but hasn't shared details on it yet. A FAQ page shows how many credits all the different actions require.

If you subscribe to the full Creative Cloud suite, you get 1,000 credits per month. However, my account let me generate only two videos, leaving over 900 credits remaining for image generation.

Firefly charges among the lowest monthly rates for a premium account among similar services. For comparison, ChatGPT requires a $20-per-month Plus account to generate videos with Sora and more than five images a day. Midjourney doesn't have a free plan at all, and its Basic Plan ($10 per month, $96 per year) lets you have just three concurrent jobs in progress and four jobs waiting in the queue at any given time. Other limitations affect how quickly the AI tool will make your artwork.

OpenArt gives you 50 image credits for free and charges $14 per month ($72 per year) for its Essential plan. Stable Diffusion has a free plan with ads, watermarks, and non-priority processing (read: slow), or you can pay $10 per month ($84 per year for a Pro account), which removes the ads and watermarks and gives you 2,000 fast image generations per month. Meta AI generates images for free if you log into a Facebook account. Which of these is the best value? It depends on what features you want, but Firefly’s combination of free functionality and affordable premium plans is appealing.

Generating Images With the Firefly Web App

Firefly on the web can generate images based on text prompts you enter, which can describe any outlandish visual idea you get. For example, you can tell it to create a realistic photo of a reindeer on a Caribbean beach with hawks flying around and a team of huskies pulling a dogsled with a driver on the shore, as I did in testing.

(Credit: Adobe/PCMag)

Image generation with AI continues to get faster and better. In testing, Firefly generated four images using the Firefly Image 3 model in just five seconds. Copilot (which uses Dall-E 3) and Gemini (which uses Imagen 3) both took 11 seconds to generate a single image. Firefly doesn't produce a full-resolution image upon the first request, however. For that, you need to tap the Upscale All button.

The way you word your text prompt is critical to how closely an image matches your expectations. Even when you think you have it perfect, you might see anomalies and distortions. But things are getting better. When I last tested Firefly with the dogsled scenario, it didn't include dogs in the results. I asked for (but didn’t get) a dogsled driver and didn't ask for (but got) beach loungers. The birds and dogs had too many limbs, and things were facing the wrong way. In my latest tests, most of those problems are gone.

Prompt Box

Midjourney and Stable Diffusion helpfully include a negative prompt box to rule out things you don’t want to see in your image. Firefly doesn’t, and when I tried giving it a negative instruction, such as “No clouds,” I still got plenty. Adding “Clouds” to Stable Diffusion’s Negative prompt box did the trick. OpenArt has a negative text box, too, but its Prompt Adherence slider, which determines how strictly the model interprets your text, is even more intriguing. Firefly doesn't have this.

(Credit: Adobe/PCMag)

After creating your image, Firefly leaves your text in the prompt box. You can only add to it if you want to tweak the same scene. One drawback of Firefly compared with Copilot, Gemini, and Meta AI is that you can't edit the image with a follow-up text prompt. With the others, I can enter a subsequent prompt that says, "Make it photorealistic," and the AIs would comply. Firefly simply produced completely new images of a photorealistic nature in this scenario.

More Controls for Making Artwork

Firefly has some controls that you don’t find in all AI image generators. Chief among them is the ability to select an aspect ratio. For example, you can choose to generate widescreen or vertical images, whereas Copilot, Gemini, and Meta AI stick to a square aspect ratio. Firefly's Aspect Ratio setting in the left panel has choices for Landscape (4:3), Portrait (3:4), Square (1:1), and Widescreen (16:9). Of the services I tested, ChatGPT Pro with Dall-E 3, Midjourney, OpenArt and Stable Diffusion let you specify different aspect ratios, including widescreen. ChatGPT/Dall-E lets you change the aspect ratio right in the chat prompt, and OpenArt has all the standard aspect ratios. Stable Diffusion’s aspect ratio choices don't cover the standard ratios like 16:9 and so on.

(Credit: Adobe/PCMag)

You can also choose to use either version 2 or 3 of Firefly. Version 3 delivers more convincing images with more detail and realism, though some may still prefer the styles that the version 2 model generates.

When you are making images in Firefly, you can choose between Art and Photo styles, as well as apply various effects. Options include Antique Photo, Bokeh, Hyper Realistic, and Pointillism. If you choose the Photo style, you get more settings specific to that image type, including Aperture, Field of View (wide-angle or telephoto, for example), and Shutter Speed. With the Art style, a good selection of mediums and textures are available in categories such as 3D, Acrylic and Oil, Architectural Sketch, Dramatic Lighting, Neon, Pencil, and Watercolor, to name just a few. But when I asked for an image in the style of Rembrandt, I got the images below, which hardly do justice to the Dutch master:

(Credit: Adobe/PCMag)

Firefly has a Suggestions slider that tries to fill out a prompt if you just enter a word or two. For example, I turned on the slider and wrote “Sun,” and the creative suggestions below popped up above the prompt box. Suggestions could be useful when you need ideas for your image.

(Credit: Adobe/PCMag)

This feature is somewhat similar to OpenArt's Enhance Prompt feature, which tries to improve your prompt using more AI, but that works with even longer and more detailed prompts.

Working With Uploaded Images

Firefly can’t create an image based only on another that you upload along with a prompt. It also can't modify an image you upload. It does, however, let you upload an image to use as a reference to base the style or the composition of the result on. You can’t blend two images as you can with Midjourney, though. Stable Diffusion has three self-explanatory options for using personal content: Copy Contour, Copy Pose, and Sketch to Image. OpenArt gives you several more ways to influence your result image based on a picture you upload. In this case, Firefly is more limited than those other services.

Similar Products

Our Current Picks for The Best AI Video Generators for 2026

Google Gemini

4.0 Excellent

Read Our Review

Grok

3.5 Good

Read Our Review

Adobe Firefly

3.5 Good

Best DealVisit Site

Buy It Now

Adobe

Visit Site

Read Our Review

Perplexity

3.5 Good

Read Our Review

Remember my not-so-successful attempt at creating an image in Rembrandt's style? After uploading an actual Rembrandt self-portrait as a reference, I got the improved result below:

(Credit: Adobe/PCMag)

It’s actually less convincing than when I tested the same prompt and reference image nine months ago, however. It's not going to convince art historians. For comparison, here's a side-by-side comparison using the same Rembrandt picture and prompt text in Firefly, ChatGPT 4, Stable Diffusion, and OpenArt:

(Credit: Adobe/OpenAI/Stability AI/OpenArt/PCMag)

Stable Diffusion performed better than Firefly and ChatGPT with Dall-E 3, but OpenArt performs this task best for my money, using the SDXL model. Meta AI doesn’t let you upload sample images.

Text Generation

You can tell Firefly to add specific text to your image in your prompt, but you might end up with a garbled mess like I did.

(Credit: Adobe/PCMag)

I asked it to add "Cheers to another great campaign" in the above example. Doing this in ChatGPT and Meta AI yielded the correct text, though Gemini botched some of the words. Midjourney generates text you put in quotation marks, but OpenArt didn't do anything with text at all. In its favor, Adobe points you to Adobe Express for adding text, a more suitable tool for that task. Still, it would do well to emulate OpenArt and simply ignore text prompts, given its performance. Serious designers should stick to dedicated tools for creating text, at least for now. Someday, it could become something that AI generators do with aplomb.

Generating Video Content

Currently a beta feature, video creation from text prompts is now possible in Firefly. It can produce video clips of just five seconds at 1080p, so don't expect to make a feature film with it. For comparison, subscribers to OpenAI Sora's Pro plan can generate video clips of up to one minute. Google's Veo 2 is also capable of generating 60-second video clips, but it's still in closed beta. Each second of Firefly video deducts 20 credits from your allotment. Strengths include animals, animation, effects like fire and smoke, and landscapes. Humans? Not so much, at this point.

You can start with text alone or upload an image that you want to start your generated video clip. You choose between widescreen (16:9) and portrait (9:16) orientations.

The results are impressively sharp, unlike the fuzzy AI-generated videos you might have seen around. You can see this clarity in the video below of its output from my prompt, "A dog swimming through a marsh with reeds and phragmites while hummingbirds fly around under a pink sky." My only disappointment with this result is that the hummingbirds are much larger than the actual birds:

It takes a while to generate a single five-second video clip, about 45 seconds in my tests. To test the photo-to-video capability, I uploaded a photo of a coworker on the roof of PCMag's NYC office and prompted it only to "Make the man fly up into the sky flapping his arms." The result was pretty good, though the AI added wings of sorts, rather than just having him flap his arms as requested:

Translation

You can now use Firefly to Translate video and audio files. In the case of the former, you can even have the translation match the lip movements in the video (lip-syncing). The Translate Audio burns up five credits per second, and Translate & Lip Sync devours 10 credits per second. After you upload a video clip and OK the warning message that you have permission to use it, Firefly generates a transcript and gives you a choice of 17 languages or regional language options (e.g., Canadian French).

A 15-second clip took about a minute to translate. The result was far from acceptable: The original untranslated voice remained, and the voice timbre didn't match the original, even though it was able to detect two different speakers. Only Enterprise subscription holders will be able to use the lip-sync capability, and my test translation was way out of sync with the speech in the video. I had better results a year ago with Wondershare Filmora's translation feature.

The Audio page of Firefly indicates that a few new dazzling features are in the works: Enhance Speech, Text to Avatar, Text to Sound Effects, and Voice to Sound Effects. You can already find some of these capabilities in video programs like CyberLink PowerDirector and Filmora.

Editing and Sharing

After Firefly generates an image, you have quite a few choices for the next steps. You can use more Adobe AI tools like Generative Fill or open it in the online version of Photoshop. Alternatively, you can have it generate another similar set of images using the same prompt or use the structure of the generated image as the basis for another new image. Surprisingly, when it comes to post-image-generation editing, Microsoft Designer does a more Adobe-like thing by highlighting objects in your created image as you hover your mouse over them. You can then blur them or punch up their colors. With Firefly, you can also send the image to Adobe Express to turn your art into a social post, complete with text and additional graphics.

(Credit: Adobe/PCMag)

With created videos, your only choice is to download the result. I would like to have at least basic trimming or splitting tools. Sora has these and lets you join multiple AI clips.

When you download or share an image you create with Firefly, Adobe applies Content Credentials metadata to indicate it is the work of AI. If you base your request on a sample you upload, the metadata indicates that as well. Adobe has long been pushing for this kind of transparency and has even established a Content Authenticity Initiative.

What Won't Firefly Do?

Firefly can produce presentable people in your results. Competitor Gemini now does this, too, though just for Advanced subscribers ($19.99 per month). You won’t, however, see anything higher than G-rated in Firefly, so don’t hope to create adult images. You can find specialized generators for those purposes, but Firefly is all about safety.

Don’t expect to include anything trademarked, either. When I asked for a group of friends drinking Coca-Cola, the resulting beverages looked more like white wine than Coke. That’s because Adobe prevents copyright infringement in Firefly. The service also has, as you might expect, a policy against creating any images that are “harmful or offensive content that could promote violence, hate speech, or other harmful activities,” including misinformation and disinformation. If you cross any of these lines (Adobe uses both machine and human reviewers to detect them), Adobe might disable your Firefly account.

Does Adobe Train Its AI With Your Content?

According to Adobe's documentation, “We don't train on any Creative Cloud subscribers’ personal content.”

That said, the company is aware that some users want to train an AI with their images to, for example, match a corporate style. It's looking into options for such people. Adobe does train on images in Adobe Stock, the company’s answer to services like Getty Images. This policy is in stark contrast with Meta AI and Google Gemini, which take advantage of their massive social networks (Facebook, Instagram, Google Photos, and YouTube) for training fodder. Both claim to train using only public images and videos, though.

Verdict: Some Bright Spots, But Room to Grow

Anyone interested in visual arts should pay attention to the quickly growing world of AI generators. No, the latest Firefly model isn't going to put most graphic designers, photographers, or video producers out of work, but it does suffice for creating whimsical visuals and has gotten better at realism (even if it's still not fully convincing). Firefly can also now generate impressive five-second video clips, too, though it's not so great at depicting humans and lacks even basic trimming and joining tools. Adobe subscribers might find Firefly helpful for their work, but competitors like OpenAI Sora and Stable Diffusion are still better at video and image generation. We haven’t, however, evaluated enough products within this category to name an Editors’ Choice winner.