Image Models
The default image model is Seedream 5.0 Pro and the default video model is seedance-2.0. You can request any other model by name; switch to GPT Image 2 or Nano Banana when you need 4K, extreme banner ratios (such as 8:1 / 1:8), or dense fine-print typography.
Seedream 5.0 ProDefault
Default image model with balanced fidelity and prompt adherence; supports text-to-image and multi-reference image-to-image (up to 10 images).
⊞Ratio
1:1 – 21:9 (8 total)Resolution
1K / 2KGPT Image 2
OpenAI's latest image model, excels at prompt adherence and text rendering. Supports text-to-image and image-to-image with multi-reference blending.
⊞Ratio
1:1 ~ 21:9 (10 options)Resolution
1K / 2K / 4KNano Banana 2
AI image generation model supporting text-to-image and image-to-image, fast and cost-effective.
⊞Ratio
1:1 ~ 8:1 (14 ratios)Resolution
1K / 2K / 4KNano Banana Pro
High-quality AI image model with superior text rendering, supports text-to-image and image-to-image.
⊞Ratio
1:1 ~ 21:9 (10 ratios)Resolution
1K / 2K / 4KMidjourney V7
Industry-leading AI image generation model, excelling at artistic styles and creative compositions, generating 4 candidate images per request.
⊞Ratio
1:1 ~ 16:9 (7 options)Output count
4 images / requestModel Comparison#
| Seedream 5.0 Pro | GPT Image 2 | Nano Banana 2 | Nano Banana Pro | Midjourney V7 | |
|---|---|---|---|---|---|
| Quality | High (production-ready) | High | High | Studio grade | Top-tier |
| Speed | Varies — see platform | 20 ~ 90s | 15 ~ 60s | 20 ~ 90s | 30 ~ 120s |
| Text rendering | Strong (~14 languages) | Accurate | Average | ~94% accuracy | Not supported |
| Ratio | 1:1 – 21:9 (8 total) | 1:1 ~ 21:9 (10 options) | 1:1 ~ 8:1 (14 ratios) | 1:1 ~ 21:9 (10 ratios) | 1:1 ~ 16:9 (7 options) |
| Resolution | 1K / 2K | 1K / 2K / 4K | 1K / 2K / 4K | 1K / 2K / 4K | Auto (no resolution option) |
| Output count | 1 | 1 | 1 | 1 | 4 images / request |
| Reference images | Up to 10 | Up to 4 | ✓ | ✓ | Up to 4 |
Seedream 5.0 Pro capabilities#
| Capability | Description |
|---|---|
| Text to image | Generate images from a text prompt |
| Image to image | Blend up to 10 reference images |
GPT Image 2 Generation#
| Capability | Description |
|---|---|
| Text-to-Image | Generate images from text prompts with strong text and detail fidelity |
| Image-to-Image | Upload up to 4 reference images; generate new images that blend the references |
Image generation capabilities#
| Capability | Description |
|---|---|
| Text-to-image | Generate images from text prompts |
| Image-to-image | Upload a reference image and generate new images based on it |
Midjourney V7#
| Capability | Description |
|---|---|
| Text-to-Image | Generate 4 candidate images from text prompts |
| Image-to-Image | Upload reference images (up to 4), generate new images based on references |
Image aspect ratios — Seedream 5.0 Pro#
1:14:33:416:99:163:22:321:9
Image aspect ratios — GPT Image 2#
1:13:22:34:33:45:44:516:99:1621:9
Image aspect ratios — Nano Banana 2#
1:12:33:23:44:34:55:49:1616:921:91:44:11:88:1
Image aspect ratios — Nano Banana Pro#
1:12:33:23:44:34:55:49:1616:921:9
Image aspect ratios — Midjourney V7#
1:14:33:416:99:163:22:3