Worth trying for anyone editing the same product shot or poster repeatedly, since it limits the drift that builds up across edits.
Kenning's take
What it does
Ideogram 4.5 is an image editing model that aims to cut the pixel shifts, colour changes and texture faults that build up when an image is edited several times. It can edit high-resolution images without shrinking them and blends an edited crop back into the original. It is available in the Ideogram app, and Ideogram's help pages also point to it for keeping a character consistent using reference images.
- Multi-turn edits with less drift
- Edits at full resolution, tested on a 24 megapixel image
- Colour, lighting and text changes
- Old photo restoration and sketch to image
- Character consistency from one to five reference images
Who it suits. Designers and marketers who edit the same image many times, such as product shots, posters and photo restorations.
Key facts
- Category
- AI models
- Part of
- Ideogram
- Made by
- Ideogram
- Launched
- 2026
- How to get it
- in the Ideogram web and iOS app; Ideogram's API also offers it, though no per-image price for 4.5 was visible on the pages fetched
- Works on
- Web, iOS
- Free plan
- Free Ideogram plan with weekly slow credits; AI editing needs priority credits from paid plans
Pricing
| Plan | Price | Billed |
|---|---|---|
| Free | $0 | month |
| Plus | $20 | month |
| Pro | $60 | month |
| Team (per user, minimum 2) | $30 | month |
FAQ
What is Ideogram 4.5 for?
Editing images over several rounds while keeping detail, colour and texture stable.
Is Ideogram 4.5 free?
Ideogram has a free plan, but its pricing page says AI editing features need priority credits, which come with paid plans from $20 a month or top-ups.
How do I keep a character consistent in Ideogram 4.5?
Ideogram's help centre says to add one to five clear reference images of the character and refer to it by its tag in the prompt.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
Ideogram 4.0
Ideogram's first open-weight text-to-image model, built for accurate text and layout in images
ACE-Step 1.5
Open-source music generation model that writes full songs with lyrics and runs on ordinary consumer GPUs.
AlphaGenome Atlas
Google DeepMind database of AI-predicted effects for every possible single-letter change in human DNA.
Amazon Nova 2
Amazon's second family of Nova AI models, covering reasoning, speech conversation and combined text-and-image generation.