Models
Image Models
Every image generation and editing model available on Kolbo.AI
Kolbo.AI gives you the leading image models from every major lab in one place, for text to image, editing, inpainting, vector graphics, and reference-driven generation.
Available Models
| Model | Provider | Modes | Resolution | Best for |
|---|---|---|---|---|
| FLUX.2 Flex | Black Forest Labs | Text to image, editing | Up to 2K | Balanced quality and cost, flexible aspect ratios |
| FLUX.2 Max | Black Forest Labs | Text to image, editing | Up to 2K | Highest fidelity in the FLUX.2 family, premium photorealism and text rendering |
| GPT Image 2 | OpenAI | Text to image, editing, no-mask inpainting, multi-image reference (up to 16) | Up to 4K (4096 x 4096) | Near-perfect text rendering, reasoning-driven intent, print-ready output, maskless edits |
| Qwen Image 2 (Standard) | Alibaba | Text to image, editing (up to 3 refs) | Multiple aspect ratios | Credit-efficient everyday generation and rapid iteration |
| Qwen Image 2 Pro | Alibaba | Text to image, editing (up to 3 refs) | Multiple aspect ratios | Maximum fidelity hero images and client deliverables |
| Nano Banana 2 | Text to image, editing, web-search grounding | Fast tier | Blazing-fast photorealism with strong prompt understanding and real-time web grounding | |
| Nano Banana 2 Pro | Text to image, editing, strong text | Up to 4K | Studio-quality 4K with accurate text rendering and pro lighting and camera controls | |
| Recraft 4 | Recraft | Text to image, editing, raster and native vector (SVG) | Vector at any scale | Logos, icons, and brand assets as clean editable vectors, plus high-fidelity raster |
| Grok Imagine Quality | xAI | Text to image, editing, Visual DNA (up to 3 refs), Cinema mode | Up to 2K | High-fidelity, detail-rich generation with precise visual accuracy across 13 aspect ratios |
| Seedream 5.0 Lite | ByteDance | Text to image, precise region editing | Multiple aspect ratios | Reasoning-driven prompt alignment, spatial understanding, surgical edits |
| Seedream 4.5 | ByteDance | Text to image, editing | Multiple aspect ratios | Sharp, prompt-accurate generation |
| Seedream 4.0 | ByteDance | Text to image, editing | Multiple aspect ratios | Reliable everyday generation |
| Higgsfield Soul | Higgsfield | Text to image | High fidelity | High-fidelity images with strong character consistency and realistic textures |
| Midjourney V8 | Midjourney | Text to image | Multiple aspect ratios | Artistic, stylized, highly aesthetic images |
| Niji V7 | Midjourney | Text to image | Multiple aspect ratios | Anime and illustration styles |
Notes
- FLUX.2 comes in two tiers: Flex for balanced cost and Max for top fidelity. Both support a separate editing mode with up to 8 reference images.
- GPT Image 2 is the strongest model for accurate in-image text and supports maskless inpainting plus up to 16 reference images.
- Nano Banana 2 has a fast tier and a Pro tier. Pro adds 4K output and reliable text rendering.
- Grok Imagine Quality supports Visual DNA for character consistency and a Cinema mode for cinematic looks.
- Recraft 4 is the model to use when you need true vector output (logos, icons) alongside raster images.
Where to use these
These models are available across the Image tools, Creative Director, Angels (multi-angle), and the Canvas Tool. See Image Tools and Creative Director for the workflows that use them.