Google Imagen 4 - Generador de Imágenes IA Avanzado

Sobre Imagen 4

Imagen 4 is Google's text-to-image model, and this page exposes three builds: Imagen 4, Imagen 4 Ultra and Imagen 4 Fast. All three work from a written prompt alone, return between one and four images per run, and accept a negative prompt, a seed and a fixed aspect ratio. The credit total is shown on the generate button before you start.

The three builds are one model family tuned to different jobs. Imagen 4 is the balanced option: strong colour and structure at a moderate wait. Ultra pushes toward photorealism and fine detail, which is the build for a hero image that will be inspected closely. Fast is the one to use while the prompt is still moving, because it returns sooner and costs the least of the three. The interface labels each tier with its own per-run price, and the generate button repeats the charge before you send anything.

There is no uploader on this page, and that is the point. Imagen 4 generates from text, so the prompt has to do all the work: subject, framing, light, lens, material. A negative prompt carries the things you do not want, and a seed reproduces a result you liked. Five aspect ratios cover the usual placements: 1:1 for feeds, 16:9 for video frames and slides, 9:16 for stories, and 3:4 or 4:3 for editorial and presentation layouts. Asking for four images is charged as a single run rather than four, so a batch of variations is the cheapest way to explore a prompt here.

Compared with the rest of the site, Imagen 4 is the still-image specialist for work that starts from language. Flux 2 is the page for edits built from several reference images, and Ideogram is the one organised around readable lettering. If you already have a product photo and need it restaged, the prompt-only workflow here will not do the job; go to Flux 2. If the picture needs a headline baked in, go to Ideogram. What Imagen 4 offers instead is a fast, uncluttered path from a description to a photographic frame, in batches, at one charge per run.

A batch of four is four variations of one prompt, not four prompts. That distinction matters because the charge covers the run: comparing two directions means two runs, and one prompt with four images only shows how the model varies a single idea. Read the four as separate takes on one description, and compare them with each other rather than with a second direction. Write for the frame you asked for, too. A 9:16 request needs a subject described as tall and central; a wide 16:9 request needs a scene with something to fill the sides.

The prompt field accepts any supported language, so a Chinese description can be sent as written, without a translation step in front of it. Keep the negative prompt short and concrete, in the same spirit as the main prompt: list the defects you keep seeing rather than a long block of prohibitions. The seed is optional and an empty field means a random draw, so fix it as soon as a frame is close, then change one word at a time and leave every other setting untouched.

Cómo funciona Imagen 4

1. 1

   Choose a build

   Imagen 4 for the balanced middle, Ultra when fidelity matters most, Fast when you are iterating. Each tier is priced separately and the price is visible on the button.

2. 2

   Write the prompt and the exclusions

   Describe subject, framing, light and material. Put unwanted details in the negative prompt, and set a seed when you want to reproduce a result instead of exploring.

3. 3

   Set the frame, the batch and the run

   Pick 1:1, 16:9, 9:16, 3:4 or 4:3, choose one to four images, and generate. The whole batch is one charge, shown before the run starts.

Para qué sirve Imagen 4

• Photoreal product shots

  Describe the object, the surface and the light, and use Ultra when the image will be used at full size.

• Editorial and hero images

  A single wide frame for an article header or landing page, generated at 16:9 without a later crop.

• Campaign exploration

  Four variations in one run let you compare directions before committing to a final render.

• Social crops

  Generate 1:1, 9:16 and 16:9 versions of the same idea for feed, story and widescreen placements.

Imagen 4 de un vistazo

Model

Google Imagen 4, Imagen 4 Ultra and Imagen 4 Fast, run through KIE

Credit cost

Shown on the generate button before each run; one charge covers the whole batch

Modes

Text to image only, no reference uploads

Aspect ratios

1:1, 16:9, 9:16, 3:4, 4:3

Images per run

1 to 4

Controls

Prompt, negative prompt, seed, aspect ratio, model tier

Preguntas frecuentes sobre Imagen 4

How much does a run cost?

The exact charge is shown on the generate button before you run, and it is charged once per run rather than once per image, so a batch of one to four costs the same. New accounts receive 300 credits at signup and can claim 50 more once a day, both expiring after 30 days; Starter is $4.90 a month and includes 300 credits.

What is the difference between Imagen 4, Ultra and Fast?

Imagen 4 balances quality and waiting time. Ultra is tuned for maximum fidelity and detail, for final images. Fast returns sooner at the lowest credit cost and is meant for ideation, not for the keeper.

Can I upload a reference image?

No. This page is text to image only. For a photo-based edit, use Flux 2 with its reference uploader, or Ideogram's masked edit when only one region needs to change.

Can I get several variations at once?

Yes, up to four images per run. Changing the seed gives a different family of results; keeping the seed and editing the prompt lets you compare one variable at a time.

Why use Imagen 4 instead of the other model pages?

It has only one input path, the prompt, and it returns photographic stills in batches. Flux 2 adds multi-reference editing and three separately tuned builds; Ideogram adds readable lettering and a mask tool. Choose Imagen 4 when there is no source image and the target is a photographic frame.

How many ideas can I compare in one run?

One. Up to four images per run are variations of the same prompt, and the charge covers the batch rather than each image. Two directions mean two runs, each with its own prompt. To keep the comparison honest, change one element at a time: with a seed set and a negative prompt that stays the same, the difference between two batches belongs to the words you changed, not to a fresh random draw.