Skip to main content

Image Model A/B

Same prompt, every image model. Pick yours.

We gave 28 image models the same prompt, one take each. Compare them side by side before you spend a credit. Not sure where to start? Read which model is best for what.

Every model below got this prompt

“A cinematic side-profile portrait of an astronaut standing on a rocky alien beach at sunset, salt spray in the air, twin suns glowing on the horizon, helmet visor catching warm orange light, weathered patches on the spacesuit. Anamorphic lens, soft rim light, sharp focus on the visor.”

Nano Banana 2 example
Image

Nano Banana 2

Multi-image fusion & editing with Google's best

Multi-image fusionNative editingStrong instruction following
FLUX 3 Image example
new
Image

FLUX 3 Image

Black Forest Labs FLUX 3: photoreal images up to 4K, edits with 10 references, web grounding

Native 4KUp to 10 reference imagesWeb-grounded generation
FLUX 3 Image Klein example
new
Image

FLUX 3 Image Klein

Fast FLUX 3 images and edits with up to 10 references, unlimited on Relax

Fast 16-step rendersUp to 10 reference imagesText rendering
MiniMax H3 Image example
new
Image

MiniMax H3 Image

Natural photoreal images up to 16 MP that follow detailed prompts, unlimited on Relax

Natural, photoreal lookFollows long, detailed promptsUp to 16 MP
Ideogram 4.5 example
new
Image

Ideogram 4.5

Ideogram 4.5: perfect text in images, natural-language edits, Precise Edit

Best-in-class text renderingNatural-language editsPrecise Edit keeps the rest
GPT Image 2.6 example
new
Image

GPT Image 2.6

OpenAI GPT Image 2.6 Sunburst & Flare, reference-faithful edits up to 4K

Reference preservationLocalized editsComplex layouts
GPT Image 2 example
Image

GPT Image 2

OpenAI GPT Image 2 with 4K rendering & sharper text

4K renderingText fidelityReference editing
Grok Image example
Image

Grok Image

Grok Imagine generation with fast edits

Fast generationNatural editsFlexible aspect ratios
FLUX.2 example
Image

FLUX.2

Pro & Dev models with multi-reference

Best-in-class anatomyPhotorealismMulti-reference support
Krea 3 example
new
Image

Krea 3

Edit and compose from up to 8 reference images, up to 4K.

Up to 8 reference imagesProduct-in-scene placementKeeps the parts you like
Krea 2 Image example
Image

Krea 2 Image

Aesthetic text-to-image in three tiers. Turbo from 0.5 credits, plus Medium + Large.

Three tiers from 0.5 creditsAesthetic by defaultStyle reference images
Riverflow example
Image

Riverflow

Fast + Pro. Text-to-image with optional reference-image guidance, up to 4K.

Reference-guided generationTwo quality tiers1K / 2K / 4K (Pro only)
Luma UNI 1 example
Image

Luma UNI 1

Standard + MAX. Text-to-image and reference-guided editing.

~3K resolutionUp to 8 referencesTwo quality tiers
Qwen Image 2 example
Image

Qwen Image 2

Multi-image fusion, text rendering & style transfer

Multi-image fusionText renderingStyle transfer
Qwen Image 2.1 example
new
Image

Qwen Image 2.1

Native 2K images, sharp text, multi-reference edits and your own LoRAs

Native 2K outputText renderingUp to 10 reference images
Qwen 3 Pro example
new
Image

Qwen 3 Pro

Next-gen image generation and editing with sharper detail

Text-to-image generationEditing with up to 3 reference imagesSharper detail
WAN 2.7 Image example
Image

WAN 2.7 Image

The newest WAN image model, in Standard and Pro tiers

Standard & Pro tiersEditing supportImage set generation
Kling Image 4 example
new
Image

Kling Image 4

Kling Image 4: native 2K to 8K images, matching image series, product pages and websites

Native 2K, 4K and 8K outputClear text in seven languagesSeries of up to 10 matching images
Kling Image example
Image

Kling Image

Kling V3 with fast edits

High-quality generationPrecise editsFlexible aspect ratios
P-Image example
Image

P-Image

Sub-second generation with text rendering

Sub-second generationNative text renderingEditing capabilities
P-Image Ideogram example
new
Image

P-Image Ideogram

Ideogram-class text rendering, priced by thinking budget

Reliable text renderingAdjustable thinking budget0.1 to 1 credit per image
Z-Image Turbo example
Image

Z-Image Turbo

6B parameter photorealistic speed demon

6B parametersPhotorealistic outputLoRA support
Seedream 4.5 example
Image

Seedream 4.5

Best character consistency, cinematic aesthetics

Hyper-realistic outputCinematic aestheticsExcellent skin texture
Seedream 5 example
new
Image

Seedream 5

Flagship Seedream 5 Pro plus the fast Flash and Lite engines

Flagship realismBest-in-class text renderingInstant results
Ernie example
Image

Ernie

Baidu's multilingual text-to-image with HD & UHD outputs

Multilingual (EN/ZH/JA)Built-in prompt expansionHD & UHD outputs
Hunyuan Image 3 example
Image

Hunyuan Image 3

Text-to-image with fast and quality modes and seven aspect presets

80B-parameter MoE architectureHigh-fidelity outputFast / Quality mode toggle
Recraft V4.1 example
Image

Recraft V4.1

Design-quality images, SVG vectors, and product mockups

Better photorealism vs V4Smoother gradients & text renderingRaster / SVG / Utility variants
PonyXL example
Image

PonyXL

Stylized generation with LoRA

Stylized artAnime excellenceExtensive LoRA ecosystem

Don't see a guide for the model you want?

Every card has a quick guide that pulls together the model's strengths, an example prompt, and the link to start generating. Hand-curated guides keep getting added. Vote on community creations to help us prioritise which model writes up next.

Visit the Community Gallery

All 52 image models and tools

Everything you can use on Pixel Dojo today, with one line on what each does best.

Image generators

  • Nano Banana 2

    Multi-image fusion & editing with Google's best

  • FLUX 3 Image

    Black Forest Labs FLUX 3: photoreal images up to 4K, edits with 10 references, web grounding

  • FLUX 3 Image Klein

    Fast FLUX 3 images and edits with up to 10 references, unlimited on Relax

  • MiniMax H3 Image

    Natural photoreal images up to 16 MP that follow detailed prompts, unlimited on Relax

  • MOSS-Image

    Detailed photos and illustrations from long prompts, with your own trained LoRAs

  • Ideogram 4.6

    Designs and photos up to native 4K, exact brand colors, see-through backgrounds and Precise Edit

  • Ideogram 4.5

    Ideogram 4.5: perfect text in images, natural-language edits, Precise Edit

  • GPT Image 2.6

    OpenAI GPT Image 2.6 Sunburst & Flare, reference-faithful edits up to 4K

  • GPT Image 2

    OpenAI GPT Image 2 with 4K rendering & sharper text

  • Grok Image

    Grok Imagine generation with fast edits

  • FLUX.2

    Pro & Dev models with multi-reference

  • Krea 3

    Edit and compose from up to 8 reference images, up to 4K.

  • Krea 2 Image

    Aesthetic text-to-image in three tiers. Turbo from 0.5 credits, plus Medium + Large.

  • Riverflow

    Fast + Pro. Text-to-image with optional reference-image guidance, up to 4K.

  • Luma UNI 1

    Standard + MAX. Text-to-image and reference-guided editing.

  • Qwen Image 2

    Multi-image fusion, text rendering & style transfer

  • Qwen Image 2.1

    Native 2K images, sharp text, multi-reference edits and your own LoRAs

  • Qwen 3 Pro

    Next-gen image generation and editing with sharper detail

  • WAN 2.6 Image

    Legacy Wan 2.6 image generation and editing

  • WAN 2.7 Image

    The newest WAN image model, in Standard and Pro tiers

  • Kling Image 4

    Kling Image 4: native 2K to 8K images, matching image series, product pages and websites

  • Kling Image

    Kling V3 with fast edits

  • WAN 2.2 Image

    Cinematic 2MP images in seconds

  • P-Image

    Sub-second generation with text rendering

  • P-Image Ideogram

    Ideogram-class text rendering, priced by thinking budget

  • Z-Image Turbo

    6B parameter photorealistic speed demon

  • Qwen Image

    Superior text rendering & precise editing

  • Seedream 4.5

    Best character consistency, cinematic aesthetics

  • Seedream 5.5

    The newest Seedream: sharp images in seconds, exact text, see-through backgrounds

  • Seedream 5

    Flagship Seedream 5 Pro plus the fast Flash and Lite engines

  • Ernie

    Baidu's multilingual text-to-image with HD & UHD outputs

  • Hunyuan Image 3

    Text-to-image with fast and quality modes and seven aspect presets

  • Muse Image

    Muse Image: Meta's image model. Make 2K images or edit with up to 10 reference photos

  • Bria Fibo 2.0

    Bria Fibo 2.0: detailed control over light, camera and color, 21 aspect ratios, 2K output and image editing

  • Recraft V4.1

    Design-quality images, SVG vectors, and product mockups

  • FLUX.1

    Advanced Flux creation with LoRA support

  • PonyXL

    Stylized generation with LoRA

  • SDXL

    Classic SDXL with LoRA support

Image editing

Characters and faces

3D

  • Rodin

    Production 3D models, fast or standard