classaiRouter

Classai Router · Image AI

Describe it.
Watch it appear.

Nano Banana Pro, GPT Image, Seedream, FLUX and Microsoft MAI in one studio. Product shots, posters and portraits up to 4K, edits from your own reference photos — paid per image with the same credits as your AI models.

Start creating API docs
image models
27
image models
resolution
up to 4K
resolution
aspect ratios
9
aspect ratios
reference images
10
reference images

From 3.3M credits per image. Failed images are free.

Prompt:

GPT Image 1.52:3 · up to 1K
Image & video models from Google OpenAI Microsoft ByteDance Black Forest Labs Alibaba MiniMax AZ8

The models

The best image models. One studio, one price list.

Pick per job: the cheapest for drafts and volume, the sharpest for the final. You see the price and typical wait before you press Generate.

Nano Banana Pro

Google

Google's best image model: sharp text on images, faithful edits, up to 4K.

Best qualityText in imagesEdit with references

from ≈ 20M credits

per image

15–40 s

up to 4K

Nano Banana 2.1

Google

The newest Nano Banana: thinks before it draws, strong edits, up to 4K at a low price.

NewEdit with referencesUp to 4K

from ≈ 5.3M credits

per image

10–25 s

up to 4K

Nano Banana 2

Google

Fast, high quality and great at edits. Up to 4K.

FastEdit with referencesUp to 4K

from ≈ 10M credits

per image

10–25 s

up to 4K

Nano Banana 2 Lite

Google

The lightest Nano Banana: quick drafts and simple edits for very few credits.

CheapestVery fast

≈ 5.3M credits

per image

5–10 s

up to 1K

Nano Banana

Google

The original Nano Banana, still loved for natural-looking edits.

Edit with references

≈ 8M credits

per image

8–15 s

up to 1K

GPT Image 2.5 Flare

OpenAI

Lowest price

OpenAI's newest image model, tuned for speed. Great prompt following and text.

NewFastText in images

from ≈ 3.3M credits

per image

10–45 s

up to 2K

GPT Image 2.5 Sunburst

OpenAI

OpenAI's newest image model, tuned for precise edits that keep everything else untouched.

NewPrecise edits

from ≈ 3.3M credits

per image

15–150 s

up to 2K

GPT Image 2

OpenAI

OpenAI's GPT Image 2: detailed posters and long prompts. High quality takes a few minutes.

PostersFollows prompts closely

from ≈ 3.3M credits

per image

20–240 s

up to 2K

GPT Image 1.5

OpenAI

OpenAI's GPT Image 1.5: reliable all-rounder for illustrations and edits.

IllustrationsEdit with references

from ≈ 3.3M credits

per image

8–60 s

up to 1K

ChatGPT Image

OpenAI

The image model used in ChatGPT, always the latest version.

ChatGPT lookEdit with references

from ≈ 3.7M credits

per image

6–60 s

up to 1K

GPT Image 1

OpenAI

OpenAI's first GPT Image model.

Classic

from ≈ 3.3M credits

per image

10–60 s

up to 1K

GPT Image 1 Mini

OpenAI

The small, very cheap GPT Image model for drafts and volume.

Very cheapFast

from ≈ 3.3M credits

per image

8–40 s

up to 1K

GPT Image 2.5 Lite

OpenAI

Budget line of GPT Image 2.5: one flat price for any size up to 4K. Slower than the official models.

BudgetSame price up to 4K

6.3M credits

per image

30–150 s

up to 4K

MAI-Image 2.6

Microsoft

Microsoft's image model: true-to-life photos with natural light and skin.

PhotorealisticEdit with references

17M credits

per image

20–40 s

up to 1K

MAI-Image 2.6 Flash

Microsoft

The faster, cheaper MAI-Image for everyday photos.

FastPhotorealistic

8.7M credits

per image

12–20 s

up to 1K

Seedream 5.0 Pro

ByteDance

Seedream's top model for complex instructions and the finest detail. Takes about 2 minutes.

Best SeedreamComplex prompts

17M credits

per image

110–140 s

up to 2K

Seedream 5.0

ByteDance

Natural Asian faces and product photos, high resolution.

Asian facesProductsUp to 3K

7.3M credits

per image

30–40 s

up to 3K

Seedream 5.0 Flash

ByteDance

Seedream's fast, low-cost version for drafts and volume.

Very cheapFast

3.7M credits

per image

15–20 s

up to 2K

Seedream 4.5

ByteDance

4K images with crisp small text: posters, ads and UI mockups.

Up to 4KText in images

8.3M credits

per image

8–20 s

up to 4K

Seedream 4.0

ByteDance

The proven Seedream: consistent results and the cheapest 4K.

Cheap 4KStable

7.3M credits

per image

7–15 s

up to 4K

FLUX.2 Max

Black Forest Labs

FLUX's flagship: the most detailed photoreal images, up to 8 reference images.

Best FLUXPhotorealistic

15M credits

per image

8–15 s

up to 2K

FLUX.2 Pro

Black Forest Labs

Photorealistic images in seconds, consistent with reference images.

PhotorealisticVery fast

6.7M credits

per image

5–10 s

up to 2K

FLUX.2 Flex

Black Forest Labs

FLUX tuned for typography: logos, signs and text-heavy designs.

Text in imagesDesign

12.7M credits

per image

15–40 s

up to 2K

FLUX.2 Klein

Black Forest Labs

The small FLUX: images in a few seconds for the lowest price.

CheapestFastest

3.3M credits

per image

2–6 s

up to 2K

FLUX.1 Dev

Black Forest Labs

The classic open FLUX model, text to image only.

ClassicText to image

4.3M credits

per image

5–8 s

up to 1K

FLUX Kontext Max

Black Forest Labs

Edits one image by instruction with the best text editing. Needs an image to edit.

Edit onlyText editing

14.3M credits

per image

8–15 s

up to 1K

FLUX Kontext Pro

Black Forest Labs

Edits one image by instruction: change the background, style or objects. Needs an image to edit.

Edit onlyFast

7.3M credits

per image

6–12 s

up to 1K

Made with Classai Router

Real outputs. Not retouched.

Every picture here came straight out of the models above. Hover to read the prompt, click to see it full size.

For AI agents & developers

Built for AI agents & developers.

Let your agent make images through MCP, call the OpenAI-compatible API from your code, or hand your coding agent one prompt that wires it into your app.

  • OpenAI-compatible
  • One key for chat, voice, image, video
  • Pay only for what comes out
  • llms.txt + OpenAPI
  • MCP server

Connect once and your agent can make images itself — in Claude Code, Claude Desktop, Cursor, opencode or any MCP client. Streamable HTTP, authenticated with your sk_live_ key.

Tools

generate_imageedit_imagelist_media_modelsstart_videoget_videospeak

Server URL

https://router.classai.id/mcp
claude mcp add --transport http classai https://router.classai.id/mcp \
  --header "Authorization: Bearer sk_live_…"

Edit with references

Change the scene. Keep the picture.

Drop in up to 10 reference images — your product, a face, a logo, a style — and say what to change. The model keeps what matters and rewrites the rest: new background, new light, new season, your label on the pack.

  • Product on any backdropStudio shots, lifestyle scenes and marketplace photos from one product picture.
  • Text that's actually readableHeadlines, prices and labels rendered sharp — even in Indonesian.
  • Same character, new sceneKeep a person or mascot consistent across a whole campaign.
After: Same scene and composition, but on a bright sunny morning with people buying coffee, dry street, blue sky
Before the edit
BeforeAfter

“Same scene and composition, but on a bright sunny morning with people buying coffee, dry street, blue sky”

Nano Banana Pro · one edit, unretouched

How it works

Three steps. No subscription.

1

Describe it

Write what you want in English or Indonesian, or start from an idea. Add up to 10 reference images to keep products, faces and logos consistent.

2

Pick a model and size

Choose the model, aspect ratio and resolution (up to 4K). The price and typical wait show before you press Generate.

3

Pay for what comes out

Credits are taken only for images that are delivered. Blocked prompts and failed generations cost nothing.

9 formats

Every shape your channel needs.

Pick the aspect ratio before you generate — the model composes for it, nothing gets cropped afterwards.

4:5 — Instagram portrait posts

Cinematic 35mm photograph of a small Indonesian warung kopi at night after rain, warm tungsten bulbs glowing, reflections on wet asphalt, a parked scooter, steam rising from glasses of kopi tubruk, shallow depth of field, Kodak Portra colors, quiet and atmospheric

Also new · Video AI

Now make it move — with sound.

Turn any picture you made into a clip with Veo, Seedance, Wan, HappyHorse, MiniMax and Oxygen: camera motion, ambience and spoken lines generated together. From 33.3M credits per video.

See Video AI
Veo 3.1 Fast · 8 s
Seedance 2.0 Fast · 5 s

Pricing

Pay per image, from 3.3M credits.

No subscription and no monthly minimum. The price per image is shown before you generate.

Failed = freeBlocked prompts, errors and timeouts are never charged.
Valid 30 daysEach credit pack lasts 30 days and works on every model — chat, voice, image and video.
Credits onlyToken packages (for coding models) don't cover images or videos — use credits.

The Pro pack alone makes about 75 images on GPT Image 2.5 Flare or 7 videos on Seedance 2.0 Mini. Images from 3.3M credits, videos from 33.3M credits.

Credit packs — the same credits work for every model

Pro

250M credits

$0.02 / 1M credits

To try every model, or for light use.

  • All active models, all API keys
  • Valid 30 days from payment

One-time payment

$4.99

Buy Pro

Max

Save 25%

500M credits

$0.015 / 1M credits

Regular daily use, cheaper per credit.

  • All active models, all API keys
  • Valid 30 days from payment

One-time payment

$7.49

Buy Max

Ultra

Best value

1 billion credits

+ Bonus 100M credits

$0.012 / 1M credits · save 40% vs Pro

The best value per credit for heavy use.

  • All active models, all API keys
  • Valid 30 days from payment

One-time payment

$11.99

Buy Ultra

FAQ

Questions, answered.

Something else? Contact our support.

No impersonation, no deepfakes. Realistic people? Say it's AI.

Can I use the images commercially?+

Yes — ads, social posts, product pages, thumbnails, courses and client work. When a picture or clip shows realistic people, tell viewers it's AI-generated.

What isn't allowed?+

Impersonating real people, deepfakes of public figures or anyone without their consent, sexual content involving minors, and content made to deceive (fake news, fake documents). The safety filter blocks prompts like these, and accounts that keep trying are closed.

What if a generation fails or the safety filter blocks it?+

You're not charged. Credits are taken only for images that are actually delivered — a blocked prompt, a timeout or an upstream error costs nothing.

What resolution can I get?+

Images: Nano Banana Pro up to 4K, Nano Banana 2.1 up to 4K, Nano Banana 2 up to 4K, Nano Banana 2 Lite up to 1K, Nano Banana up to 1K, GPT Image 2.5 Flare up to 2K, GPT Image 2.5 Sunburst up to 2K, GPT Image 2 up to 2K, GPT Image 1.5 up to 1K, ChatGPT Image up to 1K, GPT Image 1 up to 1K, GPT Image 1 Mini up to 1K, GPT Image 2.5 Lite up to 4K, MAI-Image 2.6 up to 1K, MAI-Image 2.6 Flash up to 1K, Seedream 5.0 Pro up to 2K, Seedream 5.0 up to 3K, Seedream 5.0 Flash up to 2K, Seedream 4.5 up to 4K, Seedream 4.0 up to 4K, FLUX.2 Max up to 2K, FLUX.2 Pro up to 2K, FLUX.2 Flex up to 2K, FLUX.2 Klein up to 2K, FLUX.1 Dev up to 1K, FLUX Kontext Max up to 1K, FLUX Kontext Pro up to 1K. Videos: up to 1080p, 3–30 seconds, 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 — each model shows its own options in the studio.

How long does it take?+

Images take 2–6 s on FLUX.2 Klein and 20–240 s on GPT Image 2. Videos usually take 1–10 minutes; they render in the background and wait in your Studio history.

Do I need a separate API key?+

No. The same sk_live_ key and base URL you use for AI models and voice work here. GET /v1/media/models lists every image and video model with its price. Links returned by the API last 24 hours — download what you want to keep.

Do token packages cover images and videos?+

No — images and videos are paid from credits only (token packages are for coding models). Credits are valid 30 days per purchase and also work for every AI model and voice.

Your next picture is one sentence away.

Type, pick a model, press Generate. Most images are ready before you finish your coffee.

Create a free account
Video AIVeo, Seedance, Wan, HappyHorse, MiniMax and Oxygen — clips with sound, from a sentence or a still image.Voice AINatural Indonesian text to speech and voice cloning.AI modelsGPT, Gemini, Claude, GLM, DeepSeek and more on one API key.