AI Image Generation · 6 subcategories
Which AI image generation model leads in 2026?
gpt-image-2 (medium) from OpenAI currently leads the entire image-generation field with a score of 1510, peaking in Image Editing. The 6 subcategories below cover text-to-image, editing, upscaling, in-image text rendering, virtual try-on, and selfie/avatar generation.
★
#1
Image leader
OpenAI
gpt-image-2 (medium)
1510Elo score
158Image models tracked
6Subcategories
Per-subcategory deep dive · top 10 in each lane
Six subcategories, every one fully ranked. The top three earn medal cards; the rest follow as a compact table. Open by default — no clicks needed.
Top 10 of 57
| Rank | Model | Provider | Score bar | Score | |
|---|---|---|---|---|---|
| 04 | gemini-3-pro-image-preview-2k | 1385 | — | ||
| 05 | nano-banana-pro | 1378 | Try | ||
| 06 | flux-2-max | Black Forest Labs | 1340 | Try | |
| 07 | recraft-v4 | Recraft | 1335 | Try | |
| 08 | seedream-4.5 | Bytedance | 1308 | Try | |
| 09 | flux-2-pro | Black Forest Labs | 1295 | Try | |
| 10 | hunyuan-image-3.0-instruct | Tencent | 1280 | Try |
Top 10 of 45
#03
gemini-3-pro-image-preview-2k
Google
1389
Not on Pixazo| Rank | Model | Provider | Score bar | Score | |
|---|---|---|---|---|---|
| 04 | gemini-3.1-flash-image-preview | 1387 | — | ||
| 05 | gemini-3-pro-image-preview | 1387 | — | ||
| 06 | grok-imagine-image | xAI | 1334 | Try | |
| 07 | grok-imagine-image-pro | xAI | 1316 | Try | |
| 08 | wan2.7-image-pro | Alibaba | 1304 | Try | |
| 09 | wan2.7-image | Alibaba | 1303 | Try | |
| 10 | seedream-5.0-lite | Bytedance | 1292 | Try |
Top 10 of 17
#01
Magnific Pro
Magnific
1340
Not on Pixazo#03
Krea Enhance
Krea
1290
Not on Pixazo| Rank | Model | Provider | Score bar | Score | |
|---|---|---|---|---|---|
| 04 | Magnific Standard | Magnific | 1280 | — | |
| 05 | Clarity Upscaler | Clarity | 1265 | — | |
| 06 | Real-ESRGAN x4plus | Open Source | 1240 | — | |
| 07 | SUPIR | Open Source | 1225 | — | |
| 08 | Topaz Photo AI | Topaz Labs | 1210 | Try | |
| 09 | Upscayl | Open Source | 1195 | — | |
| 10 | ESRGAN | Open Source | 1170 | — |
Top 10 of 15
| Rank | Model | Provider | Score bar | Score | |
|---|---|---|---|---|---|
| 04 | flux-2-pro | Black Forest Labs | 1310 | Try | |
| 05 | gemini-3-pro-image-preview | 1295 | — | ||
| 06 | ideogram-v2-turbo | Ideogram | 1275 | Try | |
| 07 | gpt-image-1.5-high-fidelity | OpenAI | 1265 | Try | |
| 08 | recraft-v3-mockup | Recraft | 1245 | Try | |
| 09 | seedream-4.5 | Bytedance | 1225 | Try | |
| 10 | Qwen-Image | Alibaba | 1200 | Try |
Top 10 of 11
#01
Kolors Virtual Try-On
Bytedance
1305
Not on Pixazo#03
CatVTON
Open Source
1255
Not on Pixazo| Rank | Model | Provider | Score bar | Score | |
|---|---|---|---|---|---|
| 04 | StableVITON | Open Source | 1230 | — | |
| 05 | OOTDiffusion | Open Source | 1210 | — | |
| 06 | IDM-VTON | Open Source | 1190 | — | |
| 07 | Magic Clothing | Open Source | 1170 | — | |
| 08 | StreetTryOn | Open Source | 1150 | — | |
| 09 | WildVTON | Open Source | 1130 | — | |
| 10 | Try-Diff | Open Source | 1110 | — |
Top 10 of 13
| Rank | Model | Provider | Score bar | Score | |
|---|---|---|---|---|---|
| 04 | seedream-4.5 | Bytedance | 1310 | Try | |
| 05 | PhotoMaker v2 | Tencent | 1280 | — | |
| 06 | InstantID | Open Source | 1255 | — | |
| 07 | PuLID | Open Source | 1235 | — | |
| 08 | Arc2Face | Open Source | 1210 | — | |
| 09 | IP-Adapter Face | Open Source | 1190 | — | |
| 10 | FaceShifter | Open Source | 1170 | — |
Open-weight vs proprietary
Of 158 ranked image models, 43 are open-weight or open-source. The other 115 are proprietary APIs.
The image generation master list
Top 25 unique models across all 6 image subcategories, ranked by their highest score in any subcategory.
| Rank | Model | Strongest in | Score | |
|---|---|---|---|---|
| 1 | gpt-image-2 (medium)OpenAI | Image Editing | Try | |
| 2 | gpt-image-1.5-high-fidelityOpenAI | Edit Text in Image | Try | |
| 3 | ideogram-v3-qualityIdeogram | Edit Text in Image | Try | |
| 4 | chatgpt-image-latest-high-fidelityOpenAI | Image Editing | Try | |
| 5 | gemini-3-pro-image-preview-2kGoogle | Image Editing | — | |
| 6 | gemini-3.1-flash-image-previewGoogle | Image Editing | — | |
| 7 | gemini-3-pro-image-previewGoogle | Image Editing | — | |
| 8 | nano-banana-proGoogle | Selfie & Avatar | Try | |
| 9 | flux-2-maxBlack Forest Labs | Selfie & Avatar | Try | |
| 10 | recraft-v4Recraft | Edit Text in Image | Try | |
| 11 | Magnific ProMagnific | Image Upscaling | — | |
| 12 | grok-imagine-imagexAI | Image Editing | Try | |
| 13 | grok-imagine-image-proxAI | Image Editing | Try | |
| 14 | grok-imagine-image (20260207)xAI | Image Editing | Try | |
| 15 | flux-2-proBlack Forest Labs | Edit Text in Image | Try | |
| 16 | seedream-4.5Bytedance | Selfie & Avatar | Try | |
| 17 | Topaz Gigapixel 8Topaz Labs | Image Upscaling | Try | |
| 18 | Kolors Virtual Try-OnBytedance | Virtual Try-On | — | |
| 19 | wan2.7-image-proAlibaba | Image Editing | Try | |
| 20 | wan2.7-imageAlibaba | Image Editing | Try | |
| 21 | gemini-2.5-flash-image-previewGoogle | Image Editing | — | |
| 22 | hunyuan-image-3.0-instructTencent | Image Editing | Try | |
| 23 | seedream-5.0-liteBytedance | Image Editing | Try | |
| 24 | Krea EnhanceKrea | Image Upscaling | — | |
| 25 | Magnific StandardMagnific | Image Upscaling | — | |
| 26 | Clarity UpscalerClarity | Image Upscaling | — | |
| 27 | PhotoMaker v2Tencent | Selfie & Avatar | — | |
| 28 | Outfit AnyoneAlibaba | Virtual Try-On | Try | |
| 29 | ideogram-v2-turboIdeogram | Edit Text in Image | Try | |
| 30 | recraft-v3-mockupRecraft | Edit Text in Image | Try | |
| 31 | Real-ESRGAN x4plusOpen Source | Image Upscaling | — | |
| 32 | InstantIDOpen Source | Selfie & Avatar | — | |
| 33 | PuLIDOpen Source | Selfie & Avatar | — | |
| 34 | CatVTONOpen Source | Virtual Try-On | — | |
| 35 | StableVITONOpen Source | Virtual Try-On | — | |
| 36 | SUPIROpen Source | Image Upscaling | — | |
| 37 | Qwen-ImageAlibaba | Edit Text in Image | Try | |
| 38 | OOTDiffusionOpen Source | Virtual Try-On | — | |
| 39 | Topaz Photo AITopaz Labs | Image Upscaling | Try | |
| 40 | Arc2FaceOpen Source | Selfie & Avatar | — | |
| 41 | DALL-E 3 HDOpenAI | Text to Image | — | |
| 42 | Midjourney v6.2Midjourney | Text to Image | — | |
| 43 | Stable Diffusion 3.5 LargeStability AI | Text to Image | — | |
| 44 | FLUX.1 [dev]Black Forest Labs | Text to Image | Try | |
| 45 | IDM-VTONOpen Source | Virtual Try-On | — | |
| 46 | IP-Adapter FaceOpen Source | Selfie & Avatar | — | |
| 47 | Playground v3Playground | Text to Image | — | |
| 48 | Imagen 3Google | Text to Image | — | |
| 49 | Magic ClothingOpen Source | Virtual Try-On | — | |
| 50 | FaceShifterOpen Source | Selfie & Avatar | — | |
| 51 | UpscaylOpen Source | Image Upscaling | — | |
| 52 | Adobe Firefly 3Adobe | Text to Image | — | |
| 53 | Recraft v3Recraft | Text to Image | Try | |
| 54 | Ideogram v2Ideogram | Text to Image | Try | |
| 55 | ESRGANOpen Source | Image Upscaling | — | |
| 56 | StreetTryOnOpen Source | Virtual Try-On | — | |
| 57 | WildVTONOpen Source | Virtual Try-On | — | |
| 58 | Try-DiffOpen Source | Virtual Try-On | — | |
| 59 | Hidream-I1HiDream | Text to Image | — | |
| 60 | Sana 1.6BNVIDIA | Text to Image | — | |
| 61 | PixArt-ΣOpen Source | Text to Image | — | |
| 62 | AuraFlow v0.3Open Source | Text to Image | — | |
| 63 | Lumina-NextOpen Source | Text to Image | — | |
| 64 | KolorsBytedance | Text to Image | Try | |
| 65 | Würstchen v3Open Source | Text to Image | — | |
| 66 | GLIGEN-InpaintingOpen Source | Image Editing | — | |
| 67 | OmniEditorTencent | Image Editing | Try | |
| 68 | BrushnetOpen Source | Image Editing | — | |
| 69 | Inpaint-AnythingOpen Source | Image Editing | — | |
| 70 | PowerPaint v2Open Source | Image Editing | — | |
| 71 | DiffEditOpen Source | Image Editing | — | |
| 72 | InstructPix2PixOpen Source | Image Editing | — | |
| 73 | AnyText v2Alibaba | Edit Text in Image | Try | |
| 74 | GlyphControlOpen Source | Edit Text in Image | — | |
| 75 | TextDiffuser-2Open Source | Edit Text in Image | — | |
| 76 | Stable Diffusion XLStability AI | Text to Image | — | |
| 77 | SDXL LightningStability AI | Text to Image | — | |
| 78 | Stable CascadeStability AI | Text to Image | — | |
| 79 | SD3 MediumStability AI | Text to Image | — | |
| 80 | FLUX.1 [schnell]Black Forest Labs | Text to Image | Try | |
| 81 | FLUX.1 [pro]Black Forest Labs | Text to Image | Try | |
| 82 | Midjourney v6.1Midjourney | Text to Image | — | |
| 83 | Midjourney niji 6Midjourney | Text to Image | — | |
| 84 | Imagen 2Google | Text to Image | — | |
| 85 | Imagen 4 UltraGoogle | Text to Image | — | |
| 86 | DALL-E 2OpenAI | Text to Image | — | |
| 87 | gpt-image-1OpenAI | Text to Image | Try | |
| 88 | Leonardo PhoenixLeonardo | Text to Image | — | |
| 89 | Leonardo Vision XLLeonardo | Text to Image | — | |
| 90 | Adobe Firefly 2Adobe | Text to Image | — | |
| 91 | Recraft v3 RealisticRecraft | Text to Image | Try | |
| 92 | Seedream 3Bytedance | Text to Image | Try | |
| 93 | Wan 2.5Alibaba | Text to Image | Try | |
| 94 | Qwen-Image-PlusAlibaba | Text to Image | Try | |
| 95 | Hunyuan-Image-2Tencent | Text to Image | Try | |
| 96 | Playground v2.5Playground | Text to Image | — | |
| 97 | Pony Diffusion v6Open Source | Text to Image | — | |
| 98 | Kandinsky 3Open Source | Text to Image | — | |
| 99 | DeepFloyd IFOpen Source | Text to Image | — | |
| 100 | Realistic Vision v6Open Source | Text to Image | — | |
| 101 | Juggernaut XLOpen Source | Text to Image | — | |
| 102 | DreamShaper XLOpen Source | Text to Image | — | |
| 103 | RealVisXL v4Open Source | Text to Image | — | |
| 104 | AnimagineXL 3.1Open Source | Text to Image | — | |
| 105 | CogView-4Tsinghua | Text to Image | — | |
| 106 | PixArt-alphaOpen Source | Text to Image | — | |
| 107 | Replicate SDXLReplicate | Text to Image | — | |
| 108 | NightCafe StudioNightCafe | Text to Image | — | |
| 109 | Tencent T2I-XLTencent | Text to Image | Try | |
| 110 | DALL-E 3 StandardOpenAI | Text to Image | — | |
| 111 | Stable Diffusion 1.5Open Source | Text to Image | — | |
| 112 | Adobe Firefly Image 2Adobe | Text to Image | — | |
| 113 | ControlNet InpaintOpen Source | Image Editing | — | |
| 114 | MagicEraserMagnific | Image Editing | — | |
| 115 | Cleanup.picturesCleanup | Image Editing | — | |
| 116 | ICLightOpen Source | Image Editing | — | |
| 117 | ZeroArtOpen Source | Image Editing | — | |
| 118 | Adobe Generative FillAdobe | Image Editing | — | |
| 119 | DragGANOpen Source | Image Editing | — | |
| 120 | BlendedDiffusionOpen Source | Image Editing | — | |
| 121 | Photoshop Generative EditAdobe | Image Editing | — | |
| 122 | Replace AnythingOpen Source | Image Editing | — | |
| 123 | AnydoorOpen Source | Image Editing | — | |
| 124 | RepaintOpen Source | Image Editing | — | |
| 125 | GLIDEOpen Source | Image Editing | — | |
| 126 | ImageBartOpen Source | Image Editing | — | |
| 127 | Smart EraseTopaz Labs | Image Editing | Try | |
| 128 | ZoeDepth EditOpen Source | Image Editing | — | |
| 129 | Recraft v3 EditRecraft | Image Editing | Try | |
| 130 | Krea EditKrea | Image Editing | — | |
| 131 | Adobe Firefly EditAdobe | Image Editing | — | |
| 132 | Replicate InpaintReplicate | Image Editing | — | |
| 133 | FLUX.1 FillBlack Forest Labs | Image Editing | Try | |
| 134 | Imagen EditGoogle | Image Editing | — | |
| 135 | BigjpgBigjpg | Image Upscaling | — | |
| 136 | AI Image EnlargerAI Enlarger | Image Upscaling | — | |
| 137 | LetsEnhanceLetsEnhance | Image Upscaling | — | |
| 138 | VanceAIVanceAI | Image Upscaling | — | |
| 139 | Topaz Gigapixel 7Topaz Labs | Image Upscaling | Try | |
| 140 | Waifu2xOpen Source | Image Upscaling | — | |
| 141 | BSRGANOpen Source | Image Upscaling | — | |
| 142 | TryOnDiffusionOpen Source | Virtual Try-On | — | |
| 143 | AnyDoorOpen Source | Virtual Try-On | — | |
| 144 | ClothFitOpen Source | Virtual Try-On | — | |
| 145 | Reactor Face SwapOpen Source | Selfie & Avatar | — | |
| 146 | RoopOpen Source | Selfie & Avatar | — | |
| 147 | FaceFusionOpen Source | Selfie & Avatar | — | |
| 148 | Hyperhuman RodinDeemosTech | Selfie & Avatar | — | |
| 149 | Synthesia AvatarSynthesia | Selfie & Avatar | — | |
| 150 | AnyText v1Alibaba | Edit Text in Image | Try | |
| 151 | DiffSTEOpen Source | Edit Text in Image | — | |
| 152 | PowerPaint v1Open Source | Image Editing | — | |
| 153 | Hidream-I1 LiteHiDream | Text to Image | — | |
| 154 | Sana 0.6BNVIDIA | Text to Image | — | |
| 155 | AuraFlow v0.2Open Source | Text to Image | — | |
| 156 | Lumina-T2XOpen Source | Text to Image | — | |
| 157 | PaintByExampleOpen Source | Image Editing | — | |
| 158 | DiffEdit LiteOpen Source | Image Editing | — |
How is this ai image generation ranking calculated?
Two methods are used. Text-to-image and image-editing pull live Elo from public arenas with millions of human votes. Upscaling, in-image text editing, virtual try-on, and selfie/avatar use independent editorial scoring against fixed prompt sets, calibrated to the same scale.
- Method — Public arena (2 subcategories) + editorial (4)
- Models tracked — 158 across 6 subcategories
- Sample size — 45.0M arena votes
- Pixazo's role — Access layer, not model trainer. Pixazo never appears at #1 in any ranking.
- Limitations — Elo scores reflect aggregate human preference, not absolute capability or fit for your specific use case. Editorial scoring is calibrated but inherently subjective, and sample sizes vary across subcategories. Frontier image models often shift positions within weeks of a release — always test the top 2-3 finalists against your own prompts before committing.
FAQ — AI image generation
Which AI image generation model is best in 2026?
gpt-image-2 (medium) from OpenAI leads image generation overall with a score of 1510, peaking in the image editing subcategory. Across all 6 image-generation subcategories, OpenAI, Google, and Bytedance dominate the top spots.
Are open-source AI image models competitive in 2026?
Yes. Of the 158 ranked image models, 43 are open-weight (27%). Hunyuan-Image-3.0 and Qwen-Image are competitive open alternatives to proprietary frontier models.
Which AI model is best for image editing specifically?
OpenAI's gpt-image-2 (medium) leads image editing with an Elo of 1510. ChatGPT image and Google's Gemini 3 Pro Image are close behind. For open-weight editing, Alibaba's Qwen-Image-Edit is the strongest option.
Which AI model is best for upscaling images?
Magnific Pro leads creative upscaling, while Topaz Gigapixel 8 is best for photo restoration. For open-weight self-hosting, Real-ESRGAN x4plus is the strongest free option.
Can I generate text inside images with AI?
Yes. OpenAI's gpt-image-2 (medium) and Ideogram v3 lead text-rendering inside images. Recraft v4 is the strongest option for multi-language in-image text.
Can I try these image AI models on Pixazo?
Yes. Where a 'Try' button is shown, the model is available in the Pixazo Playground or Pixazo Models API. No separate API key per provider needed.