Ideogram 4.5
Ideogram 4.5 is the model Ideogram released on 30 September 2026, four months after the open-weight 4.0. Ideogram pitches it as an editing model: its claim is that the edited image keeps everything you did not ask to change, so you can edit the same picture again and again without the artifacts and colour shifts that build up in other models. It also generates from text, with Ideogram's usual strength in set type. Unlike 4.0, its weights are not published yet; Ideogram says they are coming. On fal it is priced per image by quality tier, whatever the size.
Anvisha Pai, Co-founder & CEO, Voyager
Ideogram · released
›What we asked for
A self-portrait of Ideogram 4.5, an AI image model built for precise edits and typography, as a bold typographic poster: the word 'IDEOGRAM' set large in a heavy grotesque, the numerals '4.5' in outline beside it, a thin registration mark and a small line reading 'edit without drift' at the foot, Swiss grid layout on warm off-white with one coral accent.
ideogram/v4.5
image_size: {"width":864,"height":1152}, quality: medium, num_images: 1
returned 864 by 1152 pixels
At a glance
- Price
- $0.03, $0.06 or $0.22 an image on fal at low, medium or high quality, at any supported size. Edits add a $0.008 very-low tier.
- Sizes
- Presets or one of 36 listed sizes, from 720 by 1280 to 3328 by 1248, all at the same price. Other sizes are rejected.
- Text
- All three lines of the poster spelled correctly at every quality tier, low included. The ladder below shows them side by side.
- Repeated edits
- Three edits in a row on the same product photo left the white background almost untouched each time. The workflow tests below show every step.
- Precise edit mode
- Set edit_precision to high and Ideogram restores the pixels outside the change. A mask and up to four reference images are optional.
- Weights
- Not published yet. Ideogram says open weights are coming; 4.0's are already out under Apache 2.0.
Other Ideogram models: Ideogram 4.0. Compare all 2.
Anvisha's take
The text-to-image half is a clear step up from 4.0. Every line of type on our poster, infographic and chart was spelled right, at every quality tier, where 4.0 at Turbo produced 'SSESSIONS S'. It also over-delivers: an extra 'LONDON' tag on the poster, a tagline under the logo, a banner over the market, step descriptions nobody asked for. The chart is the warning: the six value labels are correct but the bars are not drawn to them, so April's 170 stands taller than March's 180. The edit half is the reason to use it. The benchmark edit changed the dripper's colour and added the handle and left everything else as it was, and across three successive edits our background barely moved while GPT Image 2.5 drifted with each pass. My rule: medium for most work, low is fine for drafts with text, and reach for 4.5 when you will edit the same image more than once.
Ideogram 4.5 real output: the same 12 prompts every model gets
Every image model on this site gets the same 12 prompts, so you can compare like with like. These are the pictures Ideogram 4.5 returned, one try each, nothing re-rolled or retouched. Generated on . The 12 pictures took 12 to 27 seconds each and cost $0.72 in total.
Spelling, kerning and hierarchy of three lines of set text.
›What we asked for
A concert poster that reads 'MODA SUMMER SESSIONS' in large bold type, with the subtitle 'Friday 12 June, 8pm, Roundhouse' and a smaller line 'Tickets at the door'. Risograph style, two colours, tangerine and navy, grainy ink texture.
ideogram/v4.5
image_size: {"width":864,"height":1152}, quality: medium, num_images: 1
returned 864 by 1152 pixels
Catalogue lighting, a clean seamless background and a believable contact shadow.
›What we asked for
Studio product photograph of a matte black ceramic pour-over coffee dripper on a seamless white background, soft top light, subtle contact shadow, centred, catalogue style.
ideogram/v4.5
image_size: {"width":1024,"height":1024}, quality: medium, num_images: 1
returned 1024 by 1024 pixels
A natural face, correct finger count and hands holding an object.
›What we asked for
Close-up portrait of a woman in her 40s with short grey hair, laughing, holding a red ceramic mug with both hands, natural window light, 85mm lens, shallow depth of field.
ideogram/v4.5
image_size: {"width":864,"height":1152}, quality: medium, num_images: 1
returned 864 by 1152 pixels
A flat single-colour mark with a legible wordmark and no gradients or photo texture.
›What we asked for
A flat vector logo mark for a bicycle repair shop called 'Spoke & Chain': a single-colour icon combining a chain link and a wheel spoke, with the wordmark below in a geometric sans-serif, black on white.
ideogram/v4.5
image_size: {"width":1024,"height":1024}, quality: medium, num_images: 1
returned 1024 by 1024 pixels
Many distinct objects and people in one frame without merging or duplicating them.
›What we asked for
A crowded Saturday farmers market seen from slightly above: at least twelve stalls with striped awnings, shoppers carrying tote bags, a dog on a leash, crates of oranges, a musician with a guitar, bunting overhead, late morning light.
ideogram/v4.5
image_size: {"width":1152,"height":864}, quality: medium, num_images: 1
returned 1152 by 864 pixels
A named historical style applied faithfully rather than a generic painterly look.
›What we asked for
A lighthouse on a rocky cliff during a storm, in the style of a Japanese ukiyo-e woodblock print, with visible wood grain texture, flat colour areas and a bokashi gradient sky.
ideogram/v4.5
image_size: {"width":864,"height":1152}, quality: medium, num_images: 1
returned 864 by 1152 pixels
Six labelled bars at the stated heights, in order, with readable axes.
›What we asked for
A clean bar chart titled 'Monthly active users' showing six months, January through June, with the values 120, 150, 180, 170, 220 and 260, labelled axes, white background, one accent colour, no 3D effects.
ideogram/v4.5
image_size: {"width":1152,"height":864}, quality: medium, num_images: 1
returned 1152 by 864 pixels
A 21:9 composition that leaves the left third clear for headline text.
›What we asked for
Website hero banner: a wide panorama of a sunlit modern co-working space with plants, the left third of the frame kept as clear wall space for text, warm neutral palette.
ideogram/v4.5
image_size: {"width":3072,"height":1280}, quality: medium, num_images: 1
returned 3072 by 1280 pixels
Four numbered steps in the given order with their exact wording, plus an icon each.
›What we asked for
An infographic card titled 'How to brew pour-over' with four numbered steps in this order: 1. Rinse the filter, 2. Bloom for 30 seconds, 3. Pour in slow circles, 4. Wait 3 minutes. A simple icon beside each step, muted palette, clean sans-serif type.
ideogram/v4.5
image_size: {"width":864,"height":1152}, quality: medium, num_images: 1
returned 864 by 1152 pixels
Appetising, menu-grade food with correct textures and no melted anatomy.
›What we asked for
Overhead photograph of a margherita pizza on a wooden board, fresh basil leaves, one slice pulled out with the cheese stretching, restaurant menu photography, natural light.
ideogram/v4.5
image_size: {"width":1024,"height":1024}, quality: medium, num_images: 1
returned 1024 by 1024 pixels
Exact counts and left-right placement of five named objects.
›What we asked for
A glass of water on a wooden table with exactly three ice cubes in it, a yellow pencil resting across the rim of the glass, a blue book to the left of the glass and a red apple to the right, photographed from table height.
ideogram/v4.5
image_size: {"width":1152,"height":864}, quality: medium, num_images: 1
returned 1152 by 864 pixels
Change one object's colour and add a part while leaving background, light and framing untouched.
›What we asked for
Change the coffee dripper to a deep forest green and add a small brass handle on its right side. Keep the background, lighting and composition exactly as they are. (applied to the product on white result above)
ideogram/v4.5/edit
quality: medium, edit_precision: regular, num_images: 1
returned 1024 by 1024 pixels
How the cost under each picture was worked out: fal's Ideogram 4.5 price of $0.06 per image at medium quality, size not priced (fal.ai/models/ideogram/v4.5, read 2026-10-22); fal does not return the charge.
Additional workflow tests
Separate from the common benchmark above. These tests examine longer shots, successive edits or specialized controls. One attempt per test, including failures. Costs are estimates from published rates.
Edit 1: change only color
View the generated starting image
Ideogram 4.5
Sequential editing from the same starting image; look for changes outside the requested region.
›What we asked for
Change only the black coffee dripper to deep forest green. Preserve its exact shape, position, lighting, shadow and white background.
ideogram/v4.5/edit
quality: medium, edit_precision: high, num_images: 1
returned 1024 by 1024 pixels
Generated . Cost basis: fal's Ideogram 4.5 price of $0.06 per image at medium quality, size not priced, edit_precision high; fal lists no separate price for precision (fal.ai/models/ideogram/v4.5/edit, read 2026-10-22); fal does not return the charge. Original output.
Run at high edit precision. Only the dripper changes colour; the white background and shadow are almost identical to the source. A faint pink fringe appears along the rim.
Edit 2: add a brass handle
View the generated starting image
Ideogram 4.5
Edits the previous output above.
Sequential editing from the same starting image; look for changes outside the requested region.
›What we asked for
Add one small brass handle on the right side of the forest green coffee dripper. Change nothing else.
ideogram/v4.5/edit
quality: medium, edit_precision: high, num_images: 1
returned 1024 by 1024 pixels
Generated . Cost basis: fal's Ideogram 4.5 price of $0.06 per image at medium quality, size not priced, edit_precision high; fal lists no separate price for precision (fal.ai/models/ideogram/v4.5/edit, read 2026-10-22); fal does not return the charge. Original output.
A brass handle on the right and nothing else moved: the background differs from the previous step by a fraction of a level on a 0 to 255 scale.
Edit 3: restore black
View the generated starting image
Ideogram 4.5
Edits the previous output above.
Sequential editing from the same starting image; look for changes outside the requested region.
›What we asked for
Change only the forest green ceramic to matte black. Keep the brass handle, dripper geometry, background, shadow and framing unchanged.
ideogram/v4.5/edit
quality: medium, edit_precision: high, num_images: 1
returned 1024 by 1024 pixels
Generated . Cost basis: fal's Ideogram 4.5 price of $0.06 per image at medium quality, size not priced, edit_precision high; fal lists no separate price for precision (fal.ai/models/ideogram/v4.5/edit, read 2026-10-22); fal does not return the charge. Original output.
Back to near-black with a slight green cast, handle kept. Across the three edits the background stayed closer to the source than either GPT Image 2.5 model's did on the same chain.
Identity in a new setting
View the generated starting image
Ideogram 4.5
Preserve the same person and clothing while changing the setting.
›What we asked for
Move this same person into a bright photography studio with a plain light grey background. Preserve their face, age, hairstyle, clothing, pose and the objects in their hands. Only change the background.
ideogram/v4.5/edit
quality: medium, edit_precision: high, num_images: 1
returned 896 by 1200 pixels
Generated . Cost basis: fal's Ideogram 4.5 price of $0.06 per image at medium quality, size not priced, edit_precision high; fal lists no separate price for precision (fal.ai/models/ideogram/v4.5/edit, read 2026-10-22); fal does not return the charge. Original output.
Same face, hair, shirt and mug against plain grey, but the frame pulled back to show more of her, so 'only change the background' was not followed exactly.
Identity after a second edit
Ideogram 4.5
Edits the previous output above.
Keep recognizable features through a second style change.
›What we asked for
Convert this image into a delicate watercolor illustration. Preserve the same person, facial features, hairstyle, clothing, pose and objects. Keep the studio background plain.
ideogram/v4.5/edit
quality: medium, edit_precision: high, num_images: 1
returned 896 by 1200 pixels
Generated . Cost basis: fal's Ideogram 4.5 price of $0.06 per image at medium quality, size not priced, edit_precision high; fal lists no separate price for precision (fal.ai/models/ideogram/v4.5/edit, read 2026-10-22); fal does not return the charge. Original output.
A faithful watercolour of the same person, pose and mug on a plain ground.
Compare Ideogram 4.5
Ideogram 4.5 settings that matter
Quality
Unlike 4.0's rendering speeds, quality here is a polish dial, not a spelling dial. All three posters set 'MODA SUMMER SESSIONS', the date line and 'Tickets at the door' correctly; low added a 'LONDON' tag and a busier sunburst, high gave the most refined halftone. The pizzas differ in styling more than in quality. Times barely move, 11 to 18 seconds, while the price goes from $0.03 to $0.22, so high is seven times low for a finish most people will not need.
Text in an image
›What we asked for
A concert poster that reads 'MODA SUMMER SESSIONS' in large bold type, with the subtitle 'Friday 12 June, 8pm, Roundhouse' and a smaller line 'Tickets at the door'. Risograph style, two colours, tangerine and navy, grainy ink texture.
ideogram/v4.5
image_size: {"width":864,"height":1152}, quality: low, num_images: 1
returned 864 by 1152 pixels
Food photography
›What we asked for
Overhead photograph of a margherita pizza on a wooden board, fresh basil leaves, one slice pulled out with the cheese stretching, restaurant menu photography, natural light.
ideogram/v4.5
image_size: {"width":1024,"height":1024}, quality: low, num_images: 1
returned 1024 by 1024 pixels
Ideogram 4.5 pricing
| Where | Unit | Price | Source |
|---|---|---|---|
| fal | Image, medium quality Any supported size; fal says size does not change the price. Medium is the default. | $0.06 | fal model page, read |
| fal | Image, low quality | $0.03 | fal model page, read |
| fal | Image, high quality | $0.22 | fal model page, read |
| fal | Edit, very low quality The edit endpoint adds this tier; low, medium and high cost the same as generation. fal lists no separate price for edit_precision. | $0.008 | fal model page, edit, read |
Is Ideogram 4.5 free?
Ideogram's own app has a free plan, but we could not confirm which model it runs: its pricing page refused our request on 22 October 2026. 4.5's weights are not out yet, and the API and hosts charge per image.
- •Ideogram app: Ideogram says 4.5 is live in the app; check the plan page for whether the free plan includes it.
- •Open weights: promised, not released as of 22 October 2026.
- •API and hosts: per image by quality tier.
Where to use Ideogram 4.5
- Ideogram app · Ideogram's own app.
- Ideogram API · Ideogram's developer API.
- fal · Text-to-image and edit endpoints, flat price per image by quality.
- Voyager · Runs inside Voyager, the agent for creative work, on Voyager credits on every paid plan (6 credits an image at medium), or with your own fal key.
Voyager
Run Ideogram 4.5 inside Voyager
Voyager is an agent for creative work that runs every model in this reference, Ideogram 4.5 included. Brief it, and it picks the model, makes the thing and hands back something you can edit. Create an account to get started.
What we noticed running Ideogram 4.5
- •All three lines spelled and set correctly, with an extra 'LONDON' tag and a record-and-palm motif nobody asked for. See the text in an image picture.
- •Title, months and all six value labels are correct, but the bars are not drawn to them: April's 170 is taller than March's 180, and the 120 bar reaches about 145. See the chart picture.
- •A single-colour chain-link mark with spokes and a correct wordmark, plus an invented 'BIKE REPAIR · EST. 2019' line. See the logo mark picture.
- •Dense and legible, with correct chalkboard spelling, but it added a 'RIVERSIDE FARMERS MARKET' banner and drew two dogs where we asked for one. See the busy scene picture.
- •Book left, apple right, ice in the glass, but the pencil stands in the water instead of resting across the rim. See the hard composition picture.
- •Not quite a pour-over cone: the dripper became a footed bowl with a pouring spout. See the product on white picture.
- •Green, with a brass handle on the right, and nothing else changed: spout, filter paper, feet, shadow and background are as they were. See the edit an existing image picture.
- •Prompt expansion is on by default and shows: most images arrive with copy, props or a tagline the prompt did not ask for. Turn it off when the layout must be exactly what you wrote.
- •Price is per image, so a 3072 by 1280 banner costs the same $0.06 as a 1024 square.
Ideogram 4.5 limits and API parameters
| Limit | Value |
|---|---|
| Output size | Presets or one of 36 listed sizes; anything else is rejected. We used 864 by 1152 for 3:4, 1152 by 864 for 4:3 and 3072 by 1280 (2.4:1) for the 21:9 banner, the nearest on the list. fal API reference, read |
| Images per request | Up to 8. fal API reference, read |
| Edit inputs | One source image plus up to four references, or three references and a mask. fal API reference, read |
›Every setting, for developers
fal (ideogram/v4.5)
fal API reference, read
| Parameter | Type | Values | Default |
|---|---|---|---|
| prompt | string | Any text Required. | |
| image_size | enum or object | square_hd, square, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9, or one of 36 listed {width, height} sizes Custom sizes must match the list, from 720 by 1280 up to 3328 by 1248. | square_hd |
| quality | enum | low, medium, high | medium |
| enable_prompt_expansion | boolean | true or false Ideogram rewrites the prompt before generating. | true |
| num_images | integer | 1 to 8 | 1 |
| seed | integer | Any integer |
fal (ideogram/v4.5/edit)
fal API reference, read
| Parameter | Type | Values | Default |
|---|---|---|---|
| image_url | string | URL Required. The image to edit. | |
| prompt | string | Any text Required. | |
| edit_precision | enum | regular, high high uses Precise Edit and restores unchanged pixels. | regular |
| quality | enum | very_low, low, medium, high With high precision the tiers run low to very_high. | medium |
| mask_url | string | URL Black is edited, white is kept. | |
| reference_image_urls | array | Up to 4 URLs 3 when a mask is used. | |
| image_size | enum or object | auto, a preset, or {width, height} auto keeps the source size. | auto |
What Ideogram 4.5 will not draw
Ideogram's terms prohibit generating content that is hateful or threatening, promotes violence or self-harm, harasses, is sexually explicit, is political, or is false, misleading or impersonates another person. The political clause is broader than most vendors' and matters for campaign or advocacy design.
Alternatives to Ideogram 4.5
Frequently asked questions
What is Ideogram 4.5?
Ideogram's image model released on 30 September 2026, built around precise editing: it changes what you ask for and keeps the rest, so the same image can be edited many times. It also generates from text.
How much does Ideogram 4.5 cost?
On fal, $0.03, $0.06 or $0.22 an image at low, medium or high quality, read on 22 October 2026, whatever the size. Edits cost the same, with an extra $0.008 very-low tier. In Voyager it is 6 credits an image at medium on any paid plan.
Is Ideogram 4.5 open source?
Not yet. Ideogram says the weights will follow, but as of 22 October 2026 only Ideogram 4.0's weights are published, under Apache 2.0.
Ideogram 4.5 or 4.0?
4.5 for type you will ship and for anything you will edit; it spelled our poster correctly even at its cheapest tier. 4.0 when you want to run the model yourself or need the lowest price per image.
How this page is made
Prices, limits and settings are read from the linked vendor and host pages on the dates shown. Every picture was generated by us through fal with the request shown under it, and the original files are kept unedited. A generation that failed is shown as a failure. Model and vendor names are used to identify the products; no vendor imagery appears on this page.
