Mage is the unlimited AI image and video generator — free to start, private by default, with no creative limits. Create characters, images, and high-quality videos fast. This blog covers tutorials, product updates, and behind-the-scenes notes from the team building it.
Almost nothing, on most platforms. Search for unlimited AI video and you'll find the word on half a dozen pricing pages, but follow it into the fine print and it usually means unlimited on a slower qu…
You can run Pony in a browser tab on any of the 5 generators below, with no model download and no local graphics processing unit (GPU). Two and a half years after release, Pony Diffusion V6 XL is stil…
The cheapest way to run Krea 2 without a graphics processing unit (GPU) is Mage, where unmetered generation starts at $10 a month. Krea's own studio runs the model too, and its unmetered Max tier star…
Our best video model, Cherry 2 Pro, is now available in 1080p, built for serious video creators.
We’re excited to introduce Grok Image 2, xAI’s newest image model, designed for photorealism and anime — now available with Unlimited Generations for Pro, Pro Plus, and Max members.
Automatic1111 built the interface most people still picture when they think about running Stable Diffusion themselves. Forge forked it, rebuilt the backend for speed and lower memory use, and became t…
We are excited to launch Cherry 2 Pro, our best video model yet, designed for serious video creators. Cherry 2 Pro is fully compatible with Characters, References, and Character Voices and is now avai…
We’re excited to introduce Guava 2, our newest exclusive fruit image model family, designed for photorealism and greater creative freedom. Both Guava 2 and Guava 2 Pro are now available with Unlimited…
Pika is genuinely fun. Pikaffects, Pikascenes, Pikadditions, Pikaswaps, Pikatwists, Pikaframes, and Pikaformance turn a still into something that melts, inflates, or explodes, and no platform on this …
To turn an AI character into a video, save the character first, then call it by its tag inside a video model that supports Characters. The character carries its own identity into the clip, so you desc…
One face, posting every day, recognizable in every shot. That's the whole job of an AI influencer, and it's where most generators quit on you. You make a portrait you love, prompt again, and a strange…
We are excited to launch Plum, a new advanced video model with Seedance 2 and Happy Horse quality. Plum is fully compatible with Characters, References, and Character Voices, and is now available with…
Because Sora is going away. OpenAI announced the discontinuation in March 2026, closed the Sora web and app experiences on April 26, 2026, and set the Sora application programming interface (API) to s…
Veo 3.1 is very good, and very cheap for what it is. Cinematic output with generated audio, strong prompt adherence, and access through the Gemini app or Google's Flow studio on a subscription that st…
Plenty of tools now do both image and video, so "all-in-one" has stopped being a real differentiator on its own. The catch is almost always the same, and it's easy to miss until you're a week into a p…
We’re excited to introduce Krea 2, a new image model designed for high-quality aesthetics. Now available with Unlimited Generations for all Premium members.
Until recently, building the audio for a scene meant stacking tools. You ran an AI voice generator in one app, scored the music in an AI music generator, pulled clips from an AI sound effect generator…
We're excited to expand Gems across Mage. You can now use Gems with every model, feature, and app while Unlimited Generations stay exactly the same.
If you've ever tried to build a series, an AI influencer, a comic, or anything that needs the same character in more than one image, you've hit the wall: the face changes every single time you generat…
Character consistency is keeping one AI-generated character, the same face and the same identity, steady across every image and video you make, instead of a slightly different person each time you gen…
We’re excited to introduce Mango 3, the next generation of our most popular image model. Mango 3 is now available with Unlimited Generations for Pro, Pro Plus, and Max members. Try it out and let us k…
Most AI face swap tools work fine until the photo gets interesting. Upload anything expressive and you get a blurred patch, a content warning, or a face that lands smooth and lifeless on top of the bo…
Outfit swapping is where most AI editors quit. Ask for a full clothing change and the tool either regenerates the whole person or blocks the edit outright.
Character swapping is the edit that breaks the most tools. You want to drop the same person into a new scene and keep them recognizable, and most editors hand you a stranger who only half resembles yo…
Messages is now live on the main site! It's a new way to privately share anything and create together with other Mage members. Start messaging here: https://www.mage.space/messages
Ecommerce sellers use AI to create product photos by uploading a basic shot of an item and generating studio-quality scenes, backgrounds, and lifestyle images around it, instead of booking a photograp…
We are excited to launch Cherry Mini, a faster and more affordable video model in the Cherry family — our flagship video model architecture. Cherry Mini delivers the same Cherry-style quality while op…
Small business runs on effective, eye-catching visuals: the Instagram post, the menu insert, the product shot, the promo banner. Whatever tool makes them has to spell your offer correctly, hand you ri…
Shopify sellers use AI to turn one photo of a real product into a full set of store images, clean hero shots, branded lifestyle scenes, and color variations, without booking a studio for every product…
Etsy sellers use AI to stage a design or product in a realistic scene, a print framed on a wall, a mug on a desk, a tee on a model, without booking a photographer or settling for the same stock templa…
Amazon sellers use AI to turn one photo of a real product into a full listing image set: the pure-white main image, the lifestyle scenes, the infographics, and the angle shots, without booking a studi…
We are excited to launch Melon, a new video model designed for anime and greater creative freedom. Melon is fully compatible with Characters and References and is now available with Unlimited Generati…
Runway is powerful. Gen-4.5 ranks among the top video models, and pros build real films with it. That power comes with a learning curve: credit budgeting, several model variants, and a workspace that …
Adobe Firefly built its name on one promise: commercially safe images trained on licensed content. That promise comes with a tax. Firefly's filters reject benign prompts, the output often reads as gen…
Running Stable Diffusion or Illustrious used to mean a local setup: a capable graphics processing unit (GPU), a Python environment, and a command line you babysit. In 2026 you can skip all of that and…
We're excited to launch Autocreate with Prompt Variation — a new way to create images and videos at scale. AI automatically expands your prompt into unique variations, helping you generate more conten…
Tensor.Art built its name on one thing: running community Stable Diffusion XL (SDXL) models, Pony, Illustrious, and other checkpoints with their Low-Rank Adaptation (LoRA) add-ons, online in the brows…
Short-form video rewards volume. If you're posting to TikTok, Reels, and Shorts, you need a tool that turns out vertical clips fast, keeps your character looking like the same person across every uplo…
Higgsfield earned its following. The cinematic presets and polished output made it a default for AI influencer video, and the quality holds up. The catch shows up once you scale: it's credit-metered, …
We’ve made several improvements to Cherry & Cherry Pro. Cherry is already our best video model — and now it's even better.
Most face swap tools do one thing well and fall apart the moment you push them. A tool nails a single frontal selfie, then smears on anything but a head-on angle. A free web app swaps one photo and as…
Skin generation is the detail that can make or break the immersion of AI-generated content. Not faces, not poses, skin. The waxy, poreless, airbrushed "AI skin" look is the most reliable way to tell a…
Anime typed into a generic AI image generator returns something vaguely Japanese. The eyes are close, the hair color makes sense, but the result reads as an anime filter rather than authentic anime. T…
We are excited to launch Grok Imagine, a fast, high-quality model family fully compatible with Characters and References — now available with Unlimited Generations!
ComfyUI is the most flexible node-based pipeline available, and for any creator willing to install Python, manage Compute Unified Device Architecture (CUDA) drivers, hunt down custom nodes, and rent o…
Midjourney v7 still produces the best artistic AI images on a one-prompt-at-a-time basis. The aesthetic baseline is real, the color grading is signature, and the cinematic compositions hold up. That's…
The 2024 generation of AI image models couldn't render a believable portrait. Skin came out plastic, eyes failed the close-up test, hands had wrong finger counts. In 2026, the leading models have move…
We are excited to launch Berry, an advanced video model designed for greater creative freedom and fully compatible with Characters and References — now available with Unlimited Generations for Max mem…
Most AI image platforms charge for output. Mage charges for membership. That structural difference compounds with volume: at 1,000 images a month, Mage Basic ($10) costs $0.010 per image. At 5,000, $0…
Photorealism is the most contested capability in AI video. Every frontier model in 2026 claims a version of it, and the language used to describe output ("cinematic," "film-quality," "indistinguishabl…
AI video generation crossed a usability threshold in 2026. Motion is stable, prompts are responsive, output resolution has caught up to professional baselines, and the quality gap between the stronges…
We’re excited to introduce Character Voice — a new way to give your Characters a consistent voice across AI video creations. You can now upload voice files and reuse them with Cherry and Raspberry. Th…
Hyper-realism stopped being a benchmark question in 2025. Frontier models all clear the bar on skin texture, fabric, and natural lighting in isolation. The question that matters in 2026 is which of th…
"AI cartoon generator" sounds like one category. In practice it's three. Some tools turn a real photo into a cartoon (a one-shot conversion). Some generate a brand new cartoon character from a text de…
A comic where the hero looks different on every page breaks the reading experience regardless of art quality. That single problem (character consistency across hundreds of panels) is what separates se…
Pixel art keeps shipping. Indie games keep selling on it, retro aesthetics keep moving merch, and asset shops keep pushing sprite packs by the thousands. The issue with most AI image models is that th…
We’re excited to introduce Audio References — a new way to keep voices, music, and sounds consistent across AI video creations. You can now upload audio files and reuse them with Cherry and Blueberry.…
AI video generation has reached production quality in 2026, but the privacy posture across the leading platforms varies enormously. Some default to public content galleries. Some train on user submiss…
The AI influencer market grew faster than the platforms built to serve it. Creators in 2024 were stitching together 4 or 5 different tools to handle character creation, image generation, video generat…
Lock a character once, reuse them across hundreds of generations without retraining or re-uploading. AI influencer creators have been asking for this since mainstream image generators arrived in 2022.…
The privacy posture of AI image and video tools has shifted sharply over the last year. Public default galleries leak prompts and outputs into searchable indexes, terms-of-service updates reserve broa…
Mage Assistant Is Here (Beta) We’re excited to introduce Mage Assistant — a chat interface designed to help with prompt writing and navigating Mage. You can find it in the bottom-right corner. Let’s …
Introducing Unlimited GPT Image 2 We’re excited to introduce GPT Image 2, a state-of-the-art image model — now available with Unlimited Generations for all Premium members. GPT Image 2 Model Concep…
Introducing Blueberry 2 Video We are excited to launch Blueberry 2, an advanced video model designed for greater creative freedom — now available with Unlimited Generations for Pro Plus and Max member…
Introducing Cherry, Our Best Video Model (Beta) We are excited to launch a brand-new video model, Cherry — currently in beta and available with Unlimited Generations for Pro Plus and Max members.
Introducing Unlimited Veo 3.1 Lite Video We’re excited to introduce Veo 3.1 Lite, a fast, high-quality video model — now available with Unlimited Generations for Pro Plus and Max members.
Introducing SDXL Plus: Adetailer + Hires Fix + 6K Images We’ve added a new model architecture, SDXL Plus. It’s an upgraded version of SDXL and works seamlessly with checkpoints like Illustrious, as we…
More Apps Are Here (Beta): Inpainting, Face Swap, Character Swap, and Outfit Swap We’re excited to introduce more Apps — a simple way to help you do one thing really well. The new apps include Inpain…
Introducing Apps: Image Enhancer, Video Enhancer, and Skin Enhancer We’re excited to introduce Apps — a simple way to help you do one thing really well. Our first three apps include Image Enhancer, Vi…
Introducing Unlimited Nano Banana 2 We’re excited to introduce Nano Banana 2, a state-of-the-art image model, now available with Unlimited Generations for Pro Plus and Max members.
Introducing Guava & Guava Pro Image Models We’re excited to introduce Guava, our newest exclusive fruit image model that excels at photorealism. Both Guava and Guava Pro are now available with Unlimi…
NEW: Create & Use Objects, Locations, Poses, and Outfits (Beta) We’re excited to bring a new References feature! You can now upload a single reference image and keep your objects, locations, poses, a…
Introducing Mango 3S We’re excited to introduce Mango 3S. Ahead of the full Mango 3 launch, we’re releasing a lighter, preview version of it. Mango 3S is now available with Unlimited Generations for P…
Overview Mango is Mage’s best image generation model and an exclusive model developed through a partner. Think of it as a fine-tuned version of one of the top image generation models on the market. M…
Mage has been adding exclusive fruit video models through its partners. Think of them as fine-tuned versions of the top video models on the market. Here’s a breakdown of the capabilities of Mage’s exc…
Overview You can import any compatible base AI models, LoRAs, or Textual Inversions into Mage. Here are the models you can import:
New Membership Plan: Max — $200 / month We’re excited to announce a new membership plan, Max. Max is designed for serious video creators who want to produce longer, richer, and more dynamic videos wi…
Introducing Motion Control We’re beyond excited to launch Motion Control. Use any video as a motion reference and bring your character to life. Motion Control is now available to all premium members …
NEW: Character Pages & Leaderboard Your published Characters now have dedicated pages you can share with others. Based on how often your Characters are played, original creators will also appear on th…
Publish Your Characters (Beta) You can now publish your Characters for others to use and see what they create with them. It’s a fun, collaborative way to bring your characters to life across the commu…
Z-Image Base Is Here We’re excited to introduce Z-Image Base, a non-distilled version of Z-Image Turbo. It delivers strong image quality, robust generative diversity, broad stylistic coverage, and pr…
Multi-Characters Feature Is Here (Beta) We’re excited to introduce Multi-Characters, a new feature that lets you use multiple consistent characters in the same image and video creations. This feature …
New Mage Is Live We’re excited to announce that New Mage is now live on the main site! https://www.mage.space/
Introducing Storyboard (Beta) You can now stitch together your video creations into a single, cohesive story. Storyboard lets you turn individual clips into a full visual narrative — all in one place.…
Create & Use Characters (Beta) We’re excited to bring a new Characters feature! You can now upload a single reference image and keep your character consistent across all AI image and video creations. …
Introducing Our New Video Model: Peach Max We’re kicking off the year with a major new launch! Introducing Peach Max — a brand-new video model now available with Unlimited Generations for Pro and Pro…
Introducing Our New Video Model: Blueberry Here’s our final highlight of the year! We are excited to launch a brand-new video model, Blueberry — now available with Unlimited Generations for Pro Plus m…
Holiday Preview: New Mage Design (Beta) Is Live We’re excited to share an early beta preview of our new Mage design! Try it out and let us know what you think: https://beta.mage.space/
Introducing Our New Video Model: Kiwi We are super excited to launch a brand-new video model, Kiwi — now available with Unlimited Generations for Pro and Pro Plus members. Try it out now and let us kn…
Introducing Mango V2 Even better Mango is here! Our best image AI model is now available with Unlimited Generations for Pro and Pro Plus members. Try it out now and let us know what you think:
Introducing Our New Video Model: Peach We are incredibly excited to bring a brand new video model, Peach — now available with Unlimited Generations for Pro and Pro Plus members.
Z-Image Is Here We’re excited to introduce Z-Image Turbo, a powerful and lightning-fast new image model that excels in photorealism, human anatomy, and text rendering. It’s now available with Unlimite…
Introducing Unlimited Flux 2 We’re incredibly excited to bring Flux 2 Dev with Unlimited Generations for all premium members! It’s the newest version of Flux, excelling in precision and realism.
Introducing Mango Fast Mode! Based on your feedback, we’re excited to introduce Fast Mode for Mango via Gems. It’s the same Mango model, but powered by multiple premium GPUs for dramatically faster r…
Qwen Image Edit Plus We’re excited to bring you Qwen Image Edit Plus with unlimited generations! It supports high-quality image editing with multi-image capability — insert multiple images and guide …
UI and Performance Improvements Based on our members’ feedback, we’ve made a number of UI and overall performance improvements. Here are the major updates:
Introducing Our New Image Model: Mango We are incredibly excited to bring a brand new Image model, Mango — now available with unlimited generations for Pro and Pro Plus members. Try it out now and le…
Our Current Focus Hi all! As some of you may already know, we’ve been exploring and experimenting with some new workflows for Mage. Right now, our focus is on building the next major features that wi…
Introducing Lightning Fast Unlimited Wan Video 2.1 & 2.2 We’re excited to roll out the new faster Wan Video Lightning models! Both Wan 2.1 Lightning and Wan 2.2 Lightning generate videos in under 60 …
Flux Krea Is Here We’re excited to introduce Flux Krea — now available to all premium members with unlimited generations! Flux Krea is a state-of-the-art open-weights FLUX model, designed for stunnin…
Introducing Wan Video 2.2 – Unlimited Generations! We’re thrilled to launch Wan Video 2.2, the best open-source AI video model — now available with unlimited generations for Pro and Pro Plus members!
We’re incredibly excited to bring the top external video models to Mage — now available via Gems!
Introducing Flux Kontext Dev – Unlimited Generations! We’re excited to bring Flux Kontext Dev to all premium members for unlimited generations! Flux Kontext Dev lets you edit an input image using sim…
This announcement was made on June 21, 2025
This announcement was made on June 13, 2025
This announcement was made on June 6, 2025
This announcement was made on May 23, 2025
This announcement was made on May 18, 2025
This announcement was made on May 16, 2025
This announcement was made on May 9, 2025
This announcement was made on May 4, 2025
