> ## Documentation Index
> Fetch the complete documentation index at: https://docs.imagine.art/llms.txt
> Use this file to discover all available pages before exploring further.

# Choosing the Right Image Model

> Find the best ImagineArt image model for your project — use case guidance, side-by-side comparisons, and direct links to every model page.

<div style={{background: "linear-gradient(135deg, #0d0016 0%, #18003a 55%, #08000f 100%)", borderRadius: "20px", padding: "3rem 3rem 2.5rem", marginBottom: "2.5rem", overflow: "hidden", position: "relative"}}>
  <div style={{position: "absolute", inset: "0", background: "radial-gradient(ellipse at 20% 50%, rgba(124,0,251,0.2) 0%, transparent 55%), radial-gradient(ellipse at 85% 20%, rgba(146,73,255,0.12) 0%, transparent 50%)", pointerEvents: "none"}} />

  <div style={{position: "relative"}}>
    <p style={{fontSize: "1.05rem", color: "rgba(255,255,255,0.85)", maxWidth: "620px", lineHeight: "1.7", margin: "0"}}>ImagineArt gives you access to 25 image generation models, each optimized for a different creative or professional workflow. Use this guide to find the right one for your project.</p>
  </div>
</div>

## Quick picks by use case

<CardGroup cols={4}>
  <Card title="Best overall" icon="star" href="/ai-models/image/imagineart-2-0">
    **ImagineArt 2.0** — ImagineArt's flagship proprietary model. Best photorealism, comprehension, and composition control.
  </Card>

  <Card title="Best for typography" icon="font-case" href="/ai-models/image/ideogram-v3">
    **Ideogram v3** — \~90–95% text accuracy. Best for posters, signage, and branded layouts.
  </Card>

  <Card title="Best open-source" icon="code" href="/ai-models/image/recraft-v4">
    **Recraft v4** — #1 HuggingFace Arena (ELO 1172). Only model with native SVG vector output.
  </Card>

  <Card title="Fastest generation" icon="bolt" href="/ai-models/image/nano-banana-2">
    **Nano Banana 2** — \~4× faster than Nano Banana Pro, up to 14 references, 4K output.
  </Card>
</CardGroup>

## By what you're building

<AccordionGroup>
  <Accordion title="Posters, banners, and print-ready assets">
    **Priority: resolution + composition + text rendering**

    For print-ready output where resolution and typographic accuracy both matter, [ImagineArt 1.5 Pro](/ai-models/image/imagineart-1-5-pro) is a strong choice — native 4K generation with enhanced composition control and reliable text rendering. For typography-dominant designs, [Ideogram v3](/ai-models/image/ideogram-v3) achieves \~90–95% text accuracy with 58 style presets.

    [Recraft v4 Pro](/ai-models/image/recraft-v4-pro) generates at 4-megapixel with native SVG vector output — the best option when you need editable, scalable assets alongside raster output. [Seedream v4.5](/ai-models/image/seedream-v4-5) and [Flux.2 Pro](/ai-models/image/flux-2-pro) both deliver production-grade text rendering at 4K.

    | Model                                                     | Resolution      | Text rendering    | Best for                            |
    | --------------------------------------------------------- | --------------- | ----------------- | ----------------------------------- |
    | [ImagineArt 1.5 Pro](/ai-models/image/imagineart-1-5-pro) | Native 4K       | Strong            | Posters, editorial, product visuals |
    | [Recraft v4 Pro](/ai-models/image/recraft-v4-pro)         | 4MP (2048×2048) | Excellent         | Print + SVG vector                  |
    | [Ideogram v3](/ai-models/image/ideogram-v3)               | Up to 1536px    | \~90–95%          | Typography-dominant designs         |
    | [Flux.2 Pro](/ai-models/image/flux-2-pro)                 | Up to 4MP       | Production-grade  | Branded print campaigns             |
    | [Seedream v4.5](/ai-models/image/seedream-v4-5)           | Up to 8K        | Excellent (dense) | Large-format, multi-language        |
  </Accordion>

  <Accordion title="Product photography and e-commerce">
    **Priority: consistency + accuracy + material rendering**

    [Nano Banana Pro](/ai-models/image/nano-banana-pro) (Google Gemini 3 Pro Image) is built for production product work — preserves the identity of up to 5 subjects for consistent brand asset creation, with Google Search grounding for accurate product rendering. [Nano Banana 2](/ai-models/image/nano-banana-2) handles up to 14 references at near-real-time speed for high-volume batches.

    [Seedream v4.5](/ai-models/image/seedream-v4-5) supports up to 14 references at 4K — ideal for multi-product campaign sets. [Flux.2 Max](/ai-models/image/flux-2-max) accepts up to 10 references with real-time web grounding.

    | Model                                               | Reference images | Output quality | Best for                       |
    | --------------------------------------------------- | ---------------- | -------------- | ------------------------------ |
    | [Nano Banana Pro](/ai-models/image/nano-banana-pro) | 5 subjects       | 4K             | Complex product composites     |
    | [Nano Banana 2](/ai-models/image/nano-banana-2)     | Up to 14         | 4K             | High-volume batch product      |
    | [Seedream v4.5](/ai-models/image/seedream-v4-5)     | Up to 14         | Up to 8K       | Multi-product campaign sets    |
    | [Flux.2 Max](/ai-models/image/flux-2-max)           | Up to 10         | Up to 4MP      | Premium product with grounding |
  </Accordion>

  <Accordion title="Portrait and lifestyle photography">
    **Priority: skin fidelity + lighting + anatomy**

    [ImagineArt 2.0](/ai-models/image/imagineart-2-0) is ImagineArt's highest-capability proprietary model — the best choice for premium portraits and lifestyle photography where the output needs to stand alongside actual photography. [ImagineArt 1.5 Pro](/ai-models/image/imagineart-1-5-pro) is the strong middle option at 4K native. [ImagineArt 1.5](/ai-models/image/imagineart-1-5) is cost-efficient for draft exploration.

    [Minimax Image](/ai-models/image/minimax-image) also excels at lifelike skin rendering with cinematic lighting inherited from MiniMax's video model lineage.

    | Model                                                     | Portrait quality | Speed | Cost tier |
    | --------------------------------------------------------- | ---------------- | ----- | --------- |
    | [ImagineArt 2.0](/ai-models/image/imagineart-2-0)         | Flagship         | Fast  | Premium   |
    | [ImagineArt 1.5 Pro](/ai-models/image/imagineart-1-5-pro) | Excellent        | Fast  | Standard  |
    | [Minimax Image](/ai-models/image/minimax-image)           | Excellent        | Fast  | —         |
    | [ImagineArt 1.5](/ai-models/image/imagineart-1-5)         | Good             | Fast  | Lower     |
  </Accordion>

  <Accordion title="Text and typography in images">
    **Priority: spelling accuracy + layout control**

    [Ideogram v3](/ai-models/image/ideogram-v3) is the clear leader — \~90–95% text accuracy with support for stylized lettering, complex multi-line layouts, and multilingual text in six languages. Use the DESIGN style type for layout-aware graphic work.

    [ChatGPT Image 2](/ai-models/image/chatgpt-image-2) delivers 99%+ text accuracy across multilingual scripts including CJK and Indic — the strongest option when text accuracy and language coverage both matter. [ChatGPT Image 1.5](/ai-models/image/chatgpt-image-1-5) (and the original [ChatGPT Image](/ai-models/image/chatgpt-image)) are best when your text references real-world knowledge — logos, flags, diagrams. [Recraft v4](/ai-models/image/recraft-v4) delivers production-quality text with native SVG output. [Z Image Turbo](/ai-models/image/z-image-turbo) and [Qwen Image](/ai-models/image/qwen-image) both excel at bilingual Chinese-English layouts.

    | Model                                                   | Text accuracy          | Multilingual         | Best for                                               |
    | ------------------------------------------------------- | ---------------------- | -------------------- | ------------------------------------------------------ |
    | [ChatGPT Image 2](/ai-models/image/chatgpt-image-2)     | 99%+                   | CJK, Indic, and more | Multilingual layouts, infographics, knowledge-grounded |
    | [Ideogram v3](/ai-models/image/ideogram-v3)             | \~90–95%               | 6 languages          | Posters, packaging, brand typography                   |
    | [ChatGPT Image 1.5](/ai-models/image/chatgpt-image-1-5) | Superior (dense)       | Limited              | Infographics, fast knowledge-grounded                  |
    | [Recraft v4](/ai-models/image/recraft-v4)               | Production-grade       | Limited              | Design, SVG, signage                                   |
    | [Qwen Image](/ai-models/image/qwen-image)               | Excellent (complex)    | ZH + EN              | Bilingual layouts, East Asian markets                  |
    | [Z Image Turbo](/ai-models/image/z-image-turbo)         | Excellent (lowest WER) | EN + ZH              | Bilingual advertising, fast iteration                  |
    | [Flux.2 Pro](/ai-models/image/flux-2-pro)               | Production-grade       | Limited              | Commercial campaigns with text                         |
  </Accordion>

  <Accordion title="Character consistency and brand mascots">
    **Priority: identity locking + cross-scene variation**

    [Nano Banana Pro](/ai-models/image/nano-banana-pro) preserves up to 5 distinct subject identities simultaneously — the strongest option for mascot or character consistency across scenes. [Seedream v4.5](/ai-models/image/seedream-v4-5) supports up to 14 reference images for multi-character compositions. [Flux.2 Max](/ai-models/image/flux-2-max) accepts up to 10 references.

    | Model                                               | Reference images | Consistency | Training needed |
    | --------------------------------------------------- | ---------------- | ----------- | --------------- |
    | [Nano Banana Pro](/ai-models/image/nano-banana-pro) | 5 subjects       | Excellent   | None            |
    | [Seedream v4.5](/ai-models/image/seedream-v4-5)     | Up to 14         | Excellent   | None            |
    | [Flux.2 Max](/ai-models/image/flux-2-max)           | Up to 10         | Strong      | None            |
    | [Nano Banana 2](/ai-models/image/nano-banana-2)     | Up to 14         | Good        | None            |
  </Accordion>

  <Accordion title="Cinematic, atmospheric, and concept art">
    **Priority: lighting depth + mood + stylistic range**

    [Dreamina Image 3.1](/ai-models/image/dreamina-3-1) is designed for cinematic aesthetics — rich lighting, atmospheric depth, and vibrant color palettes across a style range from photorealism to anime. [Midjourney V7](/ai-models/image/midjourney-v7) delivers richer textures and improved anatomical accuracy through its completely rebuilt architecture.

    [Flux Dev](/ai-models/image/flux-dev) offers strong artistic diversity as the most popular open-weight model for creative experimentation.

    | Model                                               | Cinematic quality | Style range         | Best for                          |
    | --------------------------------------------------- | ----------------- | ------------------- | --------------------------------- |
    | [Midjourney V7](/ai-models/image/midjourney-v7)     | Excellent         | Wide                | Artistic, textured, atmospheric   |
    | [Dreamina Image 3.1](/ai-models/image/dreamina-3-1) | Excellent         | Photo to anime      | Cinematic portraits, stylized art |
    | [Seedream v4](/ai-models/image/seedream-4)          | Strong            | Wide                | Concept boards, branded imagery   |
    | [Flux Dev](/ai-models/image/flux-dev)               | Strong            | Bold / experimental | Open-weight creative exploration  |
  </Accordion>

  <Accordion title="Marketing, editorial, and commercial campaigns">
    **Priority: photorealism + scalability + consistency**

    [ImagineArt 2.0](/ai-models/image/imagineart-2-0) is the flagship for high-stakes commercial imagery. [Seedream v4.5](/ai-models/image/seedream-v4-5) handles up to 14 references and generates at 4K in seconds — ideal for consistent multi-output campaigns. [Flux.2 Max](/ai-models/image/flux-2-max) with real-time web grounding is best when you need current-subject accuracy.

    | Model                                               | Photorealism | Speed | Consistency | Best for                         |
    | --------------------------------------------------- | ------------ | ----- | ----------- | -------------------------------- |
    | [ImagineArt 2.0](/ai-models/image/imagineart-2-0)   | Flagship     | Fast  | Excellent   | Hero shots, premium commercial   |
    | [Seedream v4.5](/ai-models/image/seedream-v4-5)     | Excellent    | 8–14s | Excellent   | Multi-output campaign sets       |
    | [Flux.2 Max](/ai-models/image/flux-2-max)           | Excellent    | 4–10s | Strong      | Grounded commercial imagery      |
    | [Nano Banana Pro](/ai-models/image/nano-banana-pro) | High         | Fast  | Excellent   | High-volume consistent campaigns |
  </Accordion>

  <Accordion title="Illustration, stylized art, and worldbuilding">
    **Priority: style control + detail + creative range**

    [Qwen Image](/ai-models/image/qwen-image) is ranked #1 on the AI Arena leaderboard for stylized illustration — exceptional for detailed concept art, scientific illustrations, and complex text-in-image layouts. [Dreamina Image 3.1](/ai-models/image/dreamina-3-1) supports styles from anime to Baroque oil painting. [Ideogram v3](/ai-models/image/ideogram-v3) has 58 style presets including Oil Painting, Cyberpunk, Art Deco, and more.

    | Model                                               | Illustration quality | Style presets | Best for                              |
    | --------------------------------------------------- | -------------------- | ------------- | ------------------------------------- |
    | [Qwen Image](/ai-models/image/qwen-image)           | Excellent            | Open-ended    | Concept art, scientific illustrations |
    | [Dreamina Image 3.1](/ai-models/image/dreamina-3-1) | Excellent            | Wide          | Anime, fantasy, cinematic stills      |
    | [Ideogram v3](/ai-models/image/ideogram-v3)         | Strong               | 58 presets    | Styled graphic design, mood art       |
    | [Flux Dev](/ai-models/image/flux-dev)               | Strong               | Experimental  | Bold creative, LoRA-based styles      |
  </Accordion>

  <Accordion title="Quick drafts and rapid iteration">
    **Priority: speed + low cost**

    [Nano Banana 2](/ai-models/image/nano-banana-2) is the fastest at \~4× the speed of Nano Banana Pro. [Seedream v5 Lite](/ai-models/image/seedream-v5-lite) generates in 3–5 seconds with intelligent prompt interpretation. [Z Image Turbo](/ai-models/image/z-image-turbo) delivers \~4× faster than FLUX with bilingual capability.

    A recommended workflow: iterate on [Seedream v5 Lite](/ai-models/image/seedream-v5-lite) or [Nano Banana 2](/ai-models/image/nano-banana-2) to validate direction, then move to [ImagineArt 1.5 Pro](/ai-models/image/imagineart-1-5-pro), [Seedream v4.5](/ai-models/image/seedream-v4-5), or [ImagineArt 2.0](/ai-models/image/imagineart-2-0) for the final output.

    | Model                                                 | Speed                 | Best for                           |
    | ----------------------------------------------------- | --------------------- | ---------------------------------- |
    | [Seedream v5 Lite](/ai-models/image/seedream-v5-lite) | 3–5 seconds           | Exploration, intelligent prompting |
    | [Nano Banana 2](/ai-models/image/nano-banana-2)       | \~4× faster than Pro  | High-volume, reference-based       |
    | [Z Image Turbo](/ai-models/image/z-image-turbo)       | \~4× faster than FLUX | Open-source, bilingual drafts      |
    | [ImagineArt 1.5](/ai-models/image/imagineart-1-5)     | Fast                  | Lower-cost photorealistic drafts   |
  </Accordion>

  <Accordion title="High-resolution for print (4MP+)">
    **Priority: pixel count + detail at scale**

    [Seedream v4.5](/ai-models/image/seedream-v4-5) supports up to 8192×8192 — the highest raw resolution available. [Flux.2 Max](/ai-models/image/flux-2-max) and [Flux 1.1 Ultra](/ai-models/image/flux-ultra) both generate at 4MP. [Recraft v4 Pro](/ai-models/image/recraft-v4-pro) delivers 4MP with native SVG vector alongside raster output.

    | Model                                                     | Max resolution  | Key advantage                    |
    | --------------------------------------------------------- | --------------- | -------------------------------- |
    | [Seedream v4.5](/ai-models/image/seedream-v4-5)           | Up to 8192×8192 | Highest raw resolution + 14 refs |
    | [Flux.2 Max](/ai-models/image/flux-2-max)                 | 4MP             | Web grounding + 10 refs          |
    | [Flux 1.1 Ultra](/ai-models/image/flux-ultra)             | 4MP             | Raw mode, architectural          |
    | [Recraft v4 Pro](/ai-models/image/recraft-v4-pro)         | 4MP             | SVG vector + raster at 4MP       |
    | [ImagineArt 1.5 Pro](/ai-models/image/imagineart-1-5-pro) | Native 4K       | Composition + text at 4K         |
  </Accordion>
</AccordionGroup>

## Full model comparison

| Model                                                     | Resolution   | Text               | Multi-ref | Speed                 | Best for                                            |
| --------------------------------------------------------- | ------------ | ------------------ | --------- | --------------------- | --------------------------------------------------- |
| [ImagineArt 2.0](/ai-models/image/imagineart-2-0)         | 2K           | Good               | —         | Fast                  | Flagship photorealism, commercial hero              |
| [ImagineArt 1.5 Pro](/ai-models/image/imagineart-1-5-pro) | Native 4K    | Strong             | —         | Fast                  | Posters, product visuals, professional              |
| [ImagineArt 1.5](/ai-models/image/imagineart-1-5)         | 2K           | Good               | —         | Fast                  | Cost-efficient photorealistic drafts                |
| [Nano Banana 2](/ai-models/image/nano-banana-2)           | Up to 4K     | Good               | 14        | Ultra-fast            | High-volume, rapid iteration                        |
| [Nano Banana Pro](/ai-models/image/nano-banana-pro)       | Up to 4K     | Excellent          | 14        | Fast                  | Complex compositions, production                    |
| [Nano Banana](/ai-models/image/nano-banana)               | Up to 2K     | Strong             | 4         | Near real-time        | Quick edits, product swaps                          |
| [Seedream v5 Lite](/ai-models/image/seedream-v5-lite)     | Up to 4K     | Good (titles)      | 14        | 3–5s                  | Rapid exploration, intelligent prompting            |
| [Seedream v4.5](/ai-models/image/seedream-v4-5)           | Native 4K    | Excellent (dense)  | 4         | 8–14s                 | Production deliverables, large-format               |
| [Seedream v4](/ai-models/image/seedream-4)                | Native 4K    | Multilingual       | 4         | Near real-time        | Commercial campaigns                                |
| [Recraft v4 Pro](/ai-models/image/recraft-v4-pro)         | View tooltip | Excellent          | —         | \~30s                 | Print design, SVG vector                            |
| [Recraft v4](/ai-models/image/recraft-v4)                 | View tooltip | Excellent          | —         | \~10s                 | Design, branding, SVG                               |
| [ChatGPT Image 2](/ai-models/image/chatgpt-image-2)       | Up to 4K     | 99%+, multilingual | 1         | Fastest               | Multilingual text, reasoning, character consistency |
| [ChatGPT Image 1.5](/ai-models/image/chatgpt-image-1-5)   | 1536×1024    | Superior (dense)   | 4         | Fast (4× v1)          | Infographics, knowledge-grounded                    |
| [ChatGPT Image](/ai-models/image/chatgpt-image)           | 1536×1024    | Best-in-class      | 4         | Up to 2 min           | Complex knowledge-grounded                          |
| [Midjourney V7](/ai-models/image/midjourney-v7)           | View tooltip | Good               | —         | Fast / Draft          | Artistic, cinematic, atmospheric                    |
| [xAI Grok Imagine](/ai-models/image/grok-imagine)         | 1K           | Good               | 3         | Fast                  | Real-entity accuracy, diverse styles                |
| [Flux.2 Max](/ai-models/image/flux-2-max)                 | View tooltip | Best-in-class      | 4         | 4–10s                 | Maximum quality, web grounding                      |
| [Flux.2 Pro](/ai-models/image/flux-2-pro)                 | View tooltip | Production-grade   | 4         | Fast                  | Commercial campaigns with text                      |
| [Flux 1.1 Ultra](/ai-models/image/flux-ultra)             | 2K           | Good               | —         | \~10s                 | High-res commercial, Raw mode                       |
| [Flux Dev](/ai-models/image/flux-dev)                     | 1K           | Decent             | —         | \~7–18s               | Concept art, open-weight creative                   |
| [Z Image Turbo](/ai-models/image/z-image-turbo)           | View tooltip | Excellent (EN+ZH)  | —         | \~4× faster than FLUX | Rapid generation, bilingual advertising             |
| [Dreamina Image 3.1](/ai-models/image/dreamina-3-1)       | 2K           | EN + ZH            | —         | Under 20s             | Cinematic portraits, stylized art                   |
| [Ideogram v3](/ai-models/image/ideogram-v3)               | 1K           | \~90–95%           | —         | Flash to Quality      | Typography, branding, posters                       |
| [Minimax Image](/ai-models/image/minimax-image)           | 1K           | Limited            | —         | Fast                  | Portraits, product shots                            |
| [Qwen Image](/ai-models/image/qwen-image)                 | 1K           | Excellent (ZH+EN)  | —         | Fast                  | Illustrations, bilingual, concept art               |