ImgAI

Changelog

Every model and feature we ship, newest first.

October 2026

  1. New model

    Wan 3.0 is here

    A new video model with native sound: text-to-video, first/last frame or reference images and video, from 480P to 1080P.

    New models

    • Wan 3.0 · Text, first/last frame or reference video, up to 30 seconds, priced per second.

    Improvements

    • Hotel Lobby AI · Rebuilt: upload two photos and they rap a call-and-response duet in the orange booth, with an original AI beat. Works with people, pets, paintings and cartoons. Try it
    Try it
  2. New feature

    Video edit is here

    Upload a video you already have and change its colours, clothes or objects with one sentence; motion and camera stay as they were.

    New features

    • Video edit · New scenario: upload a clip of up to 15 seconds, optionally with reference images for the new look. The result matches the original length; try a Draft before the Final. Try it

    New models

    • P-Video-Edit · Pruna's video-to-video model, priced per second of the original video. Try it
    Try it
  3. New model

    MAI-Image-2.6 is here

    Microsoft AI's in-house image model joins image creation in two versions, full and Flash, with its own model page and real examples.

    New models

    • MAI-Image-2.6 · Strong photorealism and in-image text, optional Bing web search, up to 5 reference images. Try it
    • MAI-Image-2.6 Flash · Faster and about half the price, with the same features. Try it
    Try it
  4. New model

    FLUX 3 is here

    Black Forest Labs' newest image model is now in image creation, with its own model page and real examples.

    New models

    • FLUX 3 · Generate and edit with one model. Up to 10 reference images at no extra cost, 1K to 4K. Try it
    Try it
  5. New model

    Nano Banana 2.1 is here

    Google's newest image model, released this October, is now in image creation.

    New models

    • Nano Banana 2.1 · Sharper text, steadier characters across edits and clean wide panoramas. 1K/2K/4K, up to 14 reference images, cheaper than Nano Banana 2 at 1K. Try it
    Try it
  6. New feature

    Voice studio upgrade: 16 preview languages and 22 tone presets

    Voice previews add 10 languages, tone presets apply in one click, and a faster, cheaper Fast voice model is now available.

    New features

    • Voice previews in 16 languages · Spanish, French, Portuguese, Italian, Arabic, Hindi, Indonesian, Thai, Vietnamese and Turkish join, so you can hear all 70 voices before choosing. Try it
    • 22 tone presets · Grouped by mood and use case, from warm and friendly to news anchor and live selling. Apply one in a click, then keep editing. Try it

    New models

    • Gemini 3.8 Flash-Lite TTS · The Fast voice model, used by default: quicker and uses fewer credits. Switch to High quality for more nuanced delivery. Try it
    Try it
  7. New model

    Veo 3.1 is here

    Google's Veo 3.1 joins video creation: 8-second clips with sound, priced per clip, up to 4K.

    New models

    • Veo 3.1 Fast · The best-value tier: text-to-video, first frame, first and last frames, and reference images. Try it
    • Veo 3.1 Lite · The most affordable tier, text-to-video only. Try it
    • Veo 3.1 Quality · The highest-quality tier for final cuts, from text or first and last frames. Try it
    Try it
  8. New model

    FLUX.2 and FLUX.1 Kontext are here

    Two Black Forest Labs families join image creation: Kontext edits images from a single sentence, FLUX.2 focuses on high-resolution text-to-image.

    New models

    • FLUX.1 Kontext Pro · Change backgrounds, objects or style with one sentence, up to 4 reference images. Try it
    • FLUX.1 Kontext Max · Higher quality for edits where detail and consistency matter. Try it
    • FLUX.2 Pro · Sharp, natural text-to-image, 1 to 4 megapixels. Try it
    • FLUX.2 Max · The top-quality FLUX.2 with the richest detail. Try it
    Try it
  9. New model

    Seedream 5.0 is here

    ByteDance's Seedream 5.0 Pro and Lite join image creation.

    New models

    • Seedream 5.0 Pro · Quality first: cinematic detail and accurate in-image text, 1K / 1.5K / 2K, up to 10 reference images. Try it
    • Seedream 5.0 Lite · One price for 2K, 3K and 4K, up to 14 reference images. Try it
    Try it

September 2026

  1. Release

    The new ImgAI is live

    This is a rebuilt, standalone version: the interface and the creation flow were redesigned from the ground up, and accounts do not carry over from the old site. Generating, reviewing results and making further edits all happen in one workspace, with no page switching; any result can be used as a reference image for the next round.

    For existing users: the new version requires a new registration. Unused credits from your old account transfer automatically within 24 hours, provided the new account uses exactly the same email address as the old one. If they have not arrived after 24 hours, email support@imgai.ai or reach us through the Contact link in the footer and we will reconcile it by hand.

    New features

    • Six image workflows · Freeform, AI photo editor, AI product shots, e-commerce detail-page sets, product sets, and virtual try-on, each with its own prompt template and reference slots.
    • AI video generation · Choose text/image-to-video, reference-based generation or first/last frames where supported. Models with audio support can generate sound with the video; review dialogue, lip sync and sound effects before use.
    • Plan preview for page sets · The layout and copy for each poster are drafted first; credits are charged only once you confirm.
    • Free quota and daily check-in · 10 free generations a day without an account; registered users earn credits from the daily check-in, with a bonus on the seventh consecutive day.
    • Asset library and works management · Filter by scenario, download, or use any result as the reference image for the next round.

    New models

    • Nano Banana 2 · Looks things up on the web while drawing, 4K output, up to 14 reference images. Try it
    • Nano Banana Pro · Google's studio-grade model for complex briefs and legible in-image text. Try it
    • Nano Banana · Fast 1K generation and editing. Try it
    • GPT Image 2.5 · Takes up to 16 reference images and renders dense small type and barcodes accurately. Try it
    • GPT Image 2 · Accurate text rendering with native 4K output. Try it
    • FLUX.1 Schnell · Free to use. A four-step distilled model, good for quick checks and batch drafts. Try it
    • MiniMax H3 · The default video model: 5/10/15 s, up to 2K, always with a soundtrack. Try it
    • Seedance 2.5 · Up to 30 seconds in one take, up to 1080p, multi-image reference and a soundtrack toggle. Try it
    • Seedance 2.0 · Seedance 2.0 supports native audio, first and last frames, and multimodal references. Check the generator for the credit cost of each resolution and duration. Try it
    The new ImgAI is live
    Try it
Start creating