Image Generation creates images from text descriptions. It has the widest provider selection of any mode.
For pricing details, see the Pricing page. Full model tables: Image Providers.
Providers (current)
| Provider | Strengths |
|---|---|
| Gemini (Google) | Nano Banana 2 / Lite / Pro · multi-image · search grounding · thinking |
| Flux (Black Forest Labs) | Klein / Pro / Flex / Max · strong prompt adherence |
| Qwen (Alibaba) | 3.0 / 3.0 Pro / 2.0 / Pro / Max / Plus · style presets · seeds |
| Wan (Alibaba) | 2.7 Image / Pro · multi-ref · up to 4K on Pro T2I |
| Seedream (BytePlus) | 5.0 Pro / Lite · photoreal · high native res |
| OpenAI | ChatGPT Image 2.0 · up to 16 refs |
| Microsoft | MAI Image 2.5 Flash / 2.5 / 2.5 Pro |
| Kling | 3.0 · 3.0 Omni (4K / multi-ref) |
| xAI (Grok) | Imagine 2.0 · Imagine · Imagine Quality |
| Recraft | Design / illustration (V4 / V4.1) |
| Ideogram | 4.0 · strong text-in-image |
| Meta | Muse Image 1.0 · fixed 2K per ratio · contact sheet |
| Krea | 2 Large / Medium / Medium Turbo |
| Imagen (Google) | Legacy — deprecated; avoid for new workflows |
See Image Providers for the full model list.
Settings
Aspect Ratio
Choose from presets or let the model decide (where supported):
- 1:1 — Square
- 16:9 / 9:16 — Landscape / portrait
- 4:3, 3:2, and other photo ratios
- 21:9 — Ultra-wide (where offered)
Quality
Higher quality means higher resolution output:
- 1K — fastest baseline
- 2K — common balanced tier
- 3K / 4K — on models that support it (e.g. Seedream Lite, Nano Banana Pro, Wan Pro, OpenAI, Kling Omni)
Seeds
- Random — different result each time
- Fixed seed — reproducible results with the same prompt
- Support varies by provider (e.g. Gemini, Qwen, Wan, Flux, Seedream support seeds; some providers do not)
Tips
- Start with Nano Banana 2 or Flux Klein for fast iteration, then step up quality
- Use references when you need identity, product, or style lock-in
- Prefer current models over Imagen for anything you will save as a preset or template