Updates

Neue Funktionen, Verbesserungen und Fixes für ChatImg.

v4.827.0

For Black Friday, memberships are 50% off the first period and credit packs are 30% off. Ideogram 4.6 is now available as an image model in the model picker, in chat and through MCP. You can also install the app from the home page, with ChatImg's own name and icon.

Neue Funktionen

  • New image model Ideogram 4.6: 35 credits at 1K, 88 at 2K and 175 at 4K. Pick it in the model picker, in chat or through MCP; edits keep the size of the original image.

Verbesserungen

  • Black Friday (through December 1): memberships are 50% off the first period and credit packs are 30% off. A countdown bar sits at the top of the page, and the discount applies at checkout automatically.
  • You can now "Install app" from the chatimg.ai home page, and it installs with ChatImg's own name and icon.

v4.825.0

If batch editing asks you to sign in midway, the batch now picks up automatically once you are back. A new MCP page explains how to connect, and "Connect an AI assistant / API" is one click away from the account menu and the sidebar. The Terms of Use now spell out commercial use of generated images.

Neue Funktionen

  • A new MCP page (chatimg.ai/mcp) walks you through connecting ChatImg to AI assistants such as Claude, ChatGPT and Grok Bot. The account menu has a new "Connect an AI assistant / API" entry, and "API" at the bottom of the sidebar is now "API / MCP".

Verbesserungen

  • Uploading images on the batch page and pressing generate while signed out asks you to sign in. Now, if you sign in within an hour, the batch continues on its own; no need to retry each image or upload again.
  • The Terms of Use now state that paid plans include a commercial license, so generated images can be used in client work, ads and product presentation. Images made with free credits are for trying the product.

v4.821.0

Seedream 5.5 images that occasionally came out as a collage of tiny panels are now suppressed.

Fehlerbehebungen

  • Seedream 5.5 images that occasionally came out as a full page of tiny panels instead of one image are now suppressed.

Seedream 5.5 is now available for images: areas you didn't ask to change stay the same across edit rounds, small text is clearer, and it can output transparent backgrounds. Turning images into Seedance 2.5 videos with a first frame or first and last frames no longer fails. MCP access is here: manage your character library and generate images straight from AI assistants like Claude and ChatGPT. The AI drone video page has 6 sample clips and a three-step guide, and the "Push in", "Fly through" and "Spiral up" quick routes now move the camera as named.

Neue Funktionen

  • Seedream 5.5 is now available for images: areas you didn't ask to change stay the same across edit rounds, small text is clearer, and it can output transparent PNGs. 1K and 2K, with up to 10 reference images for edits.
  • New MCP access (chatimg.ai/api/mcp): authorize it in AI assistants like Claude or ChatGPT to create and edit characters (including voice, tone and social accounts), browse models and prices, generate images, and check jobs and credits. Prices and charges match the web, failed jobs are refunded automatically, and images for a character pick up its reference images and land in your gallery. You can create dedicated keys on the API page.

Verbesserungen

  • The AI drone video page now has 6 sample clips, a three-step guide and an FAQ.

Fehlerbehebungen

  • Seedance 2.5 videos made from a first frame or first and last frames no longer fail to submit.
  • The drone video "Push in" and "Fly through" quick routes no longer produce the same move as "Pull back", and "Spiral up" is recognized correctly; hand-drawn routes still follow the path you draw.

New users can make their first GIF with the 50 free sign-up credits. A new storyboard page, /storyboard-generator, has 8 two-step templates: set up a scene, then change the shot size, camera angle or time of day in one click. When editing an image you can also switch the shot with one tap. Failed generations now tell you whether the content was blocked or no image came back, and retrying keeps your prompt; buying a subscription again renews your credits instead of resetting them, and automatic renewals add credits for the plan you are actually on.

Neue Funktionen

  • New users can make their first GIF with the 50 free sign-up credits (GPT Image 2.6 Flare, 1K); after that, GIFs go back to the regular price.
  • New storyboard page at /storyboard-generator: 8 two-step templates (shot size, camera angle, time of day, prop swap, 3-shot set). Start a scene from a template, then tap "Edit this step" to keep the composition and change just one thing.
  • When editing an image, a row of shot buttons now sits under the input: wide, medium, close-up, extreme close-up, low angle, high angle, over-the-shoulder, day, golden hour, night. Tapping one fills in a "keep the same scene, change only this" instruction you can add to before generating.

Fehlerbehebungen

  • Failed generations now say whether the content was blocked by moderation or the model returned no image, and retrying no longer drops your original prompt.
  • Buying a subscription again renews this period's credits instead of resetting them to zero.
  • Renewals now add credits based on your subscription plan, fixing cases where the renewal was charged but credits did not arrive.

New MiniMax H3 Image model with its own page: up to 4K, up to 16 reference images, 15 credits per image, and every sample on the page can be remixed in one click. Your sign-up trial credits can now be used on GPT Image 2.6 Flare too. If adding credits fails after payment, it is retried automatically.

Neue Funktionen

  • New MiniMax H3 Image model: up to 4K, up to 16 reference images, 15 credits per image, and sign-up trial credits work with it. There's also a MiniMax H3 Image page where every sample was made with the model; tap "Make this" to drop the prompt into the studio.

Verbesserungen

  • Sign-up trial credits can now be used on GPT Image 2.6 Flare too (GIFs and 2K / 4K still need paid credits).

Fehlerbehebungen

  • If adding credits fails after payment, it is retried automatically, so you won't pay without getting your credits.

The checkout page now opens in the language you picked on the site. Plan cards now count images at 15 credits each, and subscription cards say clearly which payment methods renew automatically.

Verbesserungen

  • When you pay from the pricing page or a paywall, the checkout page now shows in the language you picked on the site, instead of following your browser language. Languages the checkout page does not support yet (such as Traditional Chinese) still follow your browser.
  • Plan cards now count images at 15 credits each: about 100 images on Plus and about 330 on Pro, with a note that different models cost different amounts per image.
  • Subscription cards now spell out renewal by payment method: Alipay / WeChat Pay are one-time payments that do not renew; card / PayPal renew monthly or yearly and can be cancelled anytime. The Russian interface adds a line that card checkout does not accept Russian billing addresses, and Alipay / WeChat Pay QR payment works instead. Prices, credits and discounts are unchanged.

GIFs are here: with GPT Image 2.6, turn on the "GIF" switch to get an animated image. There are also four new GIF pages. New models GPT Image 2.6 Flare and GPT Image 2.6 Sunburst, at the same price as 2.5. Payments are credited to your signed-in account even if you pay with a different email.

Neue Funktionen

  • New: GPT Image 2.6 Flare and GPT Image 2.6 Sunburst, 50 credits per image, the same as 2.5. Flare is faster; Sunburst has finer detail. Both work for text-to-image and for editing with a reference image.
  • With GPT Image 2.6 selected there is a new "GIF" switch. Turn it on and it moves to 2.6 Flare and makes an animated GIF. A GIF costs twice the still image on the same model, so 100 credits on 2.6. Models without GIF support charge the still price. The copy button copies the link for GIFs, so you don't end up with only the first frame.
  • Four new GIF pages: AI GIF generator (/ai-gif-generator), sticker GIFs (/ai-sticker-gif-maker), product GIFs (/product-gif-generator) and pixel art animation (/pixel-art-animation-generator), each with real examples. A GIF launch bar at the top of the home page links to all four.

Fehlerbehebungen

  • When you pay by card or PayPal from the pricing page or the upgrade window, the order is credited to the account you are signed in with, even if you pay with a different email.

If you press Generate before signing in, the gate now says new accounts get 50 free credits, and that your text and images stay so generation continues after you sign in.

Verbesserungen

  • The sign-in gate on Generate now says: Sign in to generate — 50 free credits for new accounts. Your text and images are kept and continue automatically after you sign in.

A batch download ZIP can strip location and camera data from the files while the picture stays the same. You can also clean images you already have, in the browser, without signing in.

Neue Funktionen

  • New page at chatimg.ai/remove-ai-metadata: drop up to 30 JPEG, PNG, or WebP files. ChatImg removes EXIF, GPS, device info, XMP, IPTC, and C2PA in the browser. The picture is not recompressed, nothing is uploaded, and you do not need to sign in. A photo that is upright only because of its orientation tag is saved as a PNG so it stays upright. A watermark drawn into the pixels stays.
  • When you download a batch as a ZIP, you can strip file metadata. EXIF, GPS, XMP, and C2PA come out, and the pixels stay.

The home page is now an AI Character Studio: save a character once, then make stickers, figures, four-panel comics and new looks with it; picking the character brings its reference images along, and new images land in its album. If you write a prompt before signing in, it is no longer cleared: after you sign in it generates automatically. When credits run short for video, 3D, batch images and more, a pricing card opens right away; moderation blocks offer a one-tap retry with another model, and jobs stuck past the time limit refund their credits automatically. Uploading images no longer shows a false out-of-credits warning, and tapping video or 3D generate while signed out opens sign-in. The sign-in dialog mentions the 50 free credits on sign-up and no longer shows another product's branding.

Neue Funktionen

  • If you type a prompt or add reference images before signing in and press Generate, nothing is cleared: once you sign in it generates once automatically
  • Image, video and 3D jobs stuck in Processing past the time limit are marked as failed and the credits for that job are refunded
  • When moderation blocks a result, the message offers Retry with another model right there
  • The home page is now an "AI Character Studio": grow your own AI character from one image, with one-tap prompts for character assets like stickers and figures. The color, makeup and hair color tools move into a "Style Lab"
  • New character library: after generating, tap "Save as character". Next time, pick the character above the input box and its reference images come along automatically, and new images are saved to that character's album. Free accounts can save 1 character; paid accounts have no limit
  • New character play pages for stickers, figures and four-panel comics, and the personal color, hair color, makeup, historical costume and pet portrait pages can apply your character in one tap
  • New character album page to browse every saved image by character

Verbesserungen

  • When credits run short for video, 3D, X-ray, batch images or the image converter, a pricing card opens instead of only a toast
  • Pressing send with an empty box fills in a sample prompt
  • The sign-in dialog mentions 50 free credits on sign-up and uses ChatImg's own look, without another product's logo or QR code
  • Metal and gemstone labels on the jewelry page and the palm reading prompt no longer fall back to English in Traditional Chinese; Copy prompt on every scene page follows your interface language
  • Pricing cards list the character library benefits; hitting the character limit or tapping upgrade opens the payment dialog in place instead of leaving the page

Fehlerbehebungen

  • Uploading an image no longer shows a false "insufficient credits" warning and paywall
  • Video and 3D generate buttons are no longer greyed out for signed-out visitors; tapping them opens sign-in, and so does a blocked batch image run
  • The decorative icon in an empty history no longer opens the paywall by mistake
  • Buttons, menus and longer descriptions no longer fall back to English in German, French and other languages

Opening a generate link that carries a description now generates your first image automatically after sign-in. Reference images are compressed in the browser before upload, so large photos go up faster. ChatImg added to your phone's home screen no longer opens a 404 and now uses the ChatImg icon; password reset links from email no longer open a 404 page either.

Neue Funktionen

  • When you open a generate link that carries a description, you no longer have to click Generate again: you are asked to sign in if needed, then the first image is generated from that description automatically, only once, so refreshing will not charge you twice

Verbesserungen

  • Reference images are compressed in your browser before upload (long side capped at 4096, opaque images saved as JPEG), so large photos upload faster; if compressing makes it bigger or the browser lacks support, the original is uploaded as is

Fehlerbehebungen

  • ChatImg added to your phone's home screen no longer opens a 404 page, and it now uses the ChatImg icon
  • The "Reset password" link in the email no longer opens a 404 page; after changing your password you land back on the home page

On narrow phones, the logged-in top navigation and the home hero no longer overflow the screen. A new cutout page turns a photo into a transparent PNG in your browser, and the image stays on your device. Hair and glass can use the paid cutout, and you can place the subject on a studio, outdoor, or office background.

Neue Funktionen

  • New cutout page at chatimg.ai/embed/remove-background: drop in a JPG, PNG, or WebP up to 20MB and get a transparent PNG in the browser. The subject is not redrawn, the image is not uploaded, and you do not need to sign in. Switch to a white or solid-color background, then download the PNG
  • On desktop, a higher-detail option raises the resolution. The image still stays on your device and does not cost credits
  • Hair, glass, and translucent edges can use the paid cutout at 2 credits per image. Sign-in is required, and that image is uploaded. If it fails, the credits come back
  • After a cutout, place the subject on a studio, outdoor, or office background. Only an empty background is generated, then the original subject is pasted back, at 5 credits. If it fails, the credits come back. The paid cutout and scene backgrounds both work with the welcome credits

Fehlerbehebungen

  • On narrow phones, the logged-in top navigation now fits and the home hero no longer overflows the screen

Two new models: FLUX 3 Klein is fast, low-cost and goes up to 4K; Krea 3 is built for editing, so upload 1–4 images and describe the change.

Neue Funktionen

  • FLUX 3 Klein joins the model list: a lighter FLUX 3 that is fast and low-cost, goes up to 4K, works for both text-to-image and editing, and takes up to 10 reference images
  • Krea 3 joins the model list for image editing: upload 1–4 reference images and describe the change; 1K and 2K are supported. If no image is attached it tells you right away and refunds the credits

chatimg.ai opens in English by default again outside China. Language pages such as chatimg.ai/zh work again, and your first visit follows your browser's language. Homepage buttons, the model picker and the upload box are now announced properly by screen readers.

Verbesserungen

  • chatimg.ai: On your first visit, the site switches to your browser's language automatically, and remembers your choice once you switch languages yourself
  • Navigation icons, the sign-in button, the model picker and the upload box on the homepage now have readable names, so the page works well with screen readers

Fehlerbehebungen

  • chatimg.ai opens in English by default again for visitors outside China, instead of switching the whole interface to Chinese
  • Language versions such as chatimg.ai/zh load correctly again instead of still showing English

Personal color test and similar pages no longer crash in Safari on iPhone.

Fehlerbehebungen

  • The personal color test, color analysis, historical costume, wedding photoshoot and similar pages no longer crash when opened in Safari on iPhone

Scene pages now read in 15 languages instead of 4, including the personal color test, historical costume and custom photoshoot pages.

Verbesserungen

  • Body copy and instructions on scene pages such as the personal color test, historical costume and custom photoshoot now come in 15 languages instead of Chinese, English, Japanese and Korean only. Traditional Chinese uses Taiwan wording

The model list adds Qwen Image 3.0, Qwen Image 3.0 Pro, Qwen Image 2.1 Turbo, and Qwen Image 2.1 Pro. Eight models can now output 2K or 4K, and the model button shows exactly what this image will cost. The aspect ratio you pick now applies at standard size too, and 2:3 and 3:2 come out at their real ratio.

Neue Funktionen

  • The model list adds four Qwen Image tiers: 2.1 Turbo (8 credits, the fastest), 2.1 Pro (15 credits), 3.0 (18 credits, complex text-and-image prompts in one pass), and 3.0 Pro (25 credits, dense layouts such as newspapers and menus). All four can edit an uploaded image from one sentence. 2.1 Turbo and 2.1 Pro work with the welcome credits
  • The editor has a resolution picker: eight models can output 2K or 4K. The image and the charge use the same size, and a failed image is refunded in full. The model list and API pricing mark which models support 2K / 4K, and the API accepts a resolution as well

Verbesserungen

  • The price on the model button is what this image will actually cost, resolution included
  • If your credits only cover standard size, a note under the resolution picker says so and switches back in one tap

Fehlerbehebungen

  • At standard size, the aspect ratio you chose used to fall back to a square. It now uses the ratio you picked
  • 2:3 and 3:2 keep their ratio instead of snapping to 3:4 or 4:3

The credit total in the header stays as an ellipsis until the number has loaded, instead of flashing 0.

Fehlerbehebungen

  • Available credits in the header wait until the number has loaded. While it loads you see an ellipsis, not a flash of 0

Image generation can use Nano Banana 2.1, Seedream 5.0 Flash, FLUX 3, and Tencent Hy Image 3.5.

Neue Funktionen

  • Image generation can use Nano Banana 2.1. It is faster than Nano Banana 2, people and layout in the reference images stay steadier, up to 4K
  • Image generation can use Seedream 5.0 Flash. It is faster than Seedream 5 Lite, from 1K to 2K, with up to 10 reference images
  • Image generation can use FLUX 3. A local edit leaves the rest of the picture alone, up to 4K, and can combine up to 10 reference images into one
  • Image generation can use Tencent Hy Image 3.5. It can make a picture from text or change one from reference images, with up to 20 reference images, up to 4K

After you edit a prefilled prompt, submit no longer sends the example text or the example reference image with it.

Fehlerbehebungen

  • If you change a prompt that a page filled in for you, submit no longer appends that example text, and it no longer sends the example reference image. What goes out is the sentence in the box. A reference image you uploaded this time stays. A style or a purpose you already picked still goes with it

A short disconnect, or while the site is updating, no longer signs you out.

Fehlerbehebungen

  • A short disconnect or a site update no longer counts as a signed-out session, and it does not clear the login on this device

Homepage titles and CTAs now say “chat to make an image” in each language, not Generator.

Verbesserungen

  • Homepage title and H1 use local phrasing (chat to edit a photo / 画像生成AI / 이미지 생성), not Generator

The picker adds GPT Image 2.5 Flare and Sunburst. Purpose templates fill prompts without generating, Sketch handles layout, and you can pin comments to edit a spot.

Neue Funktionen

  • GPT Image 2.5 Flare (faster everyday generation and reference edits) and Sunburst (higher-precision creative generation and fine edits) are now selectable
  • Homepage Templates fill a prompt when you tap a card — posters, product shots, headshots, and more — without generating yet. Share a prompt-template link with the idea only, never someone else's photo
  • Type @Sketch or tap Sketch to open the canvas. The sketch is layout; your text is appearance. Confirming the canvas only attaches the sketch
  • Click a spot on a generated image, write a comment, and send an edit for that pin

Multi-model AI images and video: pick a model, export 4K, commercial license included.

Neue Funktionen

  • GPT Image, Nano Banana Pro, Seedream, and more in one picker
  • Image-to-video, batch generation, and a prompt gallery
  • 4K output with commercial-license plans