The AI image generator agent makes an image from a sentence or fixes a photo you already have, up to 4K.
New images or edits to your photos — 20 credits at 1K, 30 at 2K, 45 at 4K.
20 credits a picture, whether it's new or an edit.
The free image editor — and 15 credits a run once the free ones are used.
Google's quick Gemini images — blend several photos in one edit, 1K–4K.
OpenAI's image model. Strong in-image text and layout, 1K–4K.
OpenAI, 8 Sep 2026 release. Flare when you want it fast, Sunburst when an edit must not spill over; both from 25 credits.
Seedream at its best, photo-real in eight frame shapes, with in-image text in more than ten languages.
Small print stays legible: infographics, posters and storyboards, 1K or 2K.
xAI's Quality Mode sets headlines on posters and follows up to five of your images per run.
Every image model you have heard people rave about sits in one place: the SayMaker AI image generator agent. The generator above runs every image model on the site: GPT Image 2.5 in its Flare and Sunburst tiers, Nano Banana 2 and Nano Banana 2 Lite, Seedream 5.0 Lite and Pro, Qwen Image 3, Grok Imagine Image 2.0 and SayMaker Image v1, plus Nano Banana Pro, GPT Image 2, Seedream 4.0 and Seedream 4.5 for even more looks.
Each one wins a different job, and the agent picks for you when you would rather not. Nano Banana 2 Lite edits a photo for a flat 15 credits at any size. GPT Image 2.5 Sunburst is built for edits that must leave the rest of the picture alone. Qwen Image 3 keeps small text legible on infographics, Grok Imagine sets headlines on posters, and Seedream 5.0 Pro renders in-image copy in more than ten languages.
You can choose a model yourself, or describe the image to the AI image generator agent. It takes the whole visual job: it picks a model that fits the job and your account, writes a detailed prompt, and when the image should move, it plans the video step with its credits too.
Images and videos share one credit balance, and every image shows its price for the chosen model and size before it runs. Every result stays ready in your library, so a picture made today can be edited again tomorrow or become the first frame of a clip without being uploaded twice.
Turn one sentence into a finished image at up to 4K on most of the 13 image models.
Upload a photo and just type the change: background, object, outfit or style. No masks, no layers, no Photoshop.
Up to 9 photos in one run put a real person, a real product and a real setting into a single frame, and keep that face or product the same across images.
Posters, labels and infographics whose words survive the design, on models built for legible type.
The photo editor free accounts start on. Upload one photo, describe the change, and pay a flat 15 credits at 1K, 2K or 4K.
Google's Gemini image model. Combine several references in one run, 20 credits at 1K, 30 at 2K and 45 at 4K.
OpenAI's newest model and the default on a paid account. Flare for speed, Sunburst for edits, both 25 credits at 1K.
Alibaba's model for pages full of words, legible down to small captions, 25 credits at 1K and 45 at 2K.
xAI's model for designed layouts with up to five references, 45 credits at 1K.
ByteDance's flagship for photoreal portraits and in-image copy in more than ten languages.
SayMaker's own image model, at 5 credits an image, so the sign-up credits alone make four. It turns a prompt into a first picture in moments, and the idea can be pushed further on any other model in the same generator.
Choose text to image or an edit, or describe the result to the AI image generator agent.
Describe subject, style and composition; for edits, upload the photo and name the change.
Choose the size, up to 4K on most models, and the aspect ratio; the credits update with each choice.
Take the finished image, make it unmistakably yours with another edit, or give it motion instead of leaving the idea in a single frame.
The SayMaker AI image generator agent is an image generator plus an agent that can run it. Here you pick the model and settings; in the AI image generator agent you describe the image, and it chooses the model, writes the prompt and can carry the result into a video.
GPT Image 2.5 Flare and Sunburst, Nano Banana 2 and 2 Lite, Seedream 5.0 Lite and Pro, Qwen Image 3, Grok Imagine Image 2.0 and SayMaker Image v1, plus older models such as Nano Banana Pro, GPT Image 2 and Seedream 4.5 kept in the picker.
GPT Image 2.5 Flare on a paid account and Nano Banana 2 Lite on a free account. You can name any other model.
Yes. Upload it and describe the change. Nano Banana 2 Lite does it on free credits; GPT Image 2.5 Sunburst is built for edits that must leave everything else untouched.
1K, 2K and 4K on most image models; SayMaker Image v1 has a single size. Every size shows its credits up front, so the picture gets exactly the resolution it deserves.
Per image, by model and size. Some models charge one flat price at every size, such as Nano Banana 2 Lite at 15 credits; others rise with resolution. A technical failure is refunded.
Yes. New accounts get free credits for Nano Banana 2 Lite and SayMaker Image v1, and daily check-ins add more. Premium models need a plan or credit pack.
Yes. Send it to a video model from the result, or tell the AI image generator agent to animate it; the video step arrives as a plan with its credits.
Use the generator when you know the model and size. Use the AI image generator agent when the job has steps, or when you would rather describe the result than choose models.
Name the subject, the setting, the light, the framing and the style in plain sentences. For an edit, say what should change and what must stay the same. Put any text that should appear in the image inside quotation marks.
Upload a clear photo of it as a reference each time and say it must stay unchanged. Describe the new scene separately, so the AI image generator agent and the model know which parts are fixed and which it may invent.
New images can be 1:1, 16:9, 9:16, 4:3 or 3:4, and some models offer more. Edits keep the shape of the photo you upload, so your framing stays yours.
Credits are shared across everything SayMaker runs. You see the cost before each run, and a technical failure is refunded.
Checkout runs on Stripe over an encrypted connection. Your card details are handled by Stripe and are never stored by SayMaker.
Describe the result and the agent picks the model, writes the prompt and plans the video step. New accounts start on free credits, and a technical failure refunds the run.
Make my first image
