AI Image Generation Models
19 AI models that can generate images from text prompts. This includes 19 dedicated image generators and 0 text models with inline image output capabilities.
Dedicated Image Generators
(19)Purpose-built AI models designed specifically for image generation. These accept text prompts (and sometimes reference images) and produce high-quality visual output.
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 Image MiniOpenAI | 89 |
| 2 | GPT-5 ImageOpenAI | 89 |
| 3 | GPT-5.4 Image 2OpenAI | 88 |
| 4 | Nano Banana (Gemini 2.5 Flash Image)Google | 77 |
| 5 | Nano Banana 2.1Google | 40 |
| 6 | Nano Banana 2 (Gemini 3.1 Flash Image)Google | 40 |
| 7 | Nano Banana Pro (Gemini 3 Pro Image)Google | 40 |
| 8 | Nano Banana 2 (Gemini 3.1 Flash Image Preview)Google | 40 |
| 9 | Nano Banana Pro (Gemini 3 Pro Image Preview)Google | 40 |
| 10 | Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)Google | 22 |
| 11 | Midjourney v6.1Midjourney | 8 |
| 12 | DALL-E 3OpenAI | 8 |
| 13 | Stable Diffusion 3.5Stability AI | 8 |
| 14 | FLUX.1 ProBlack Forest Labs | 8 |
| 15 | Ideogram 2.0Ideogram | 8 |
| 16 | Recraft V3Recraft | 8 |
| 17 | Imagen 3Google | 8 |
| 18 | Adobe Firefly 3Adobe | 8 |
| 19 | Leonardo PhoenixLeonardo AI | 8 |
Text Models with Image Output
(0)Large language models that can generate images inline within a conversation. These combine text understanding with image creation in a single unified model.
No text models with image output available.
Dedicated vs. Inline Image Generation
Dedicated Image Generators
Purpose-built models like DALL-E, Stable Diffusion, and Flux that specialize in converting text descriptions into images. They typically produce higher-quality, more controllable visual output and support features like inpainting, outpainting, and style control. These models are optimized purely for image synthesis.
Text Models with Image Output
Large language models like GPT-4o and Gemini that can generate images as part of a text conversation. These unified models understand context, follow complex multi-step instructions, and can create images that are tightly integrated with textual reasoning. They trade some image quality for conversational flexibility.
When to Use Dedicated Generators
Choose dedicated image models when you need the highest visual quality, fine-grained control over style and composition, batch generation, or specialized features like image editing, variations, and upscaling. They are ideal for design workflows, marketing assets, and creative production pipelines.
When to Use Inline Image Models
Choose text models with image output when you need images generated as part of a larger conversation, when the image needs to reflect complex contextual understanding, or when you want a single API call that handles both text and image responses. These are great for chatbots, content creation assistants, and agentic workflows.
Pricing Comparison
How image generation models compare on cost. Prices are per million tokens for API access.
| Model | Input/1M | Output/1M |
|---|---|---|
| Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) | $0.25 | $1.50 |
| GPT-5 Image Mini | $2.50 | $2.00 |
| Nano Banana (Gemini 2.5 Flash Image) | $0.30 | $2.50 |
| Nano Banana 2 (Gemini 3.1 Flash Image) | $0.50 | $3.00 |
| Nano Banana 2 (Gemini 3.1 Flash Image Preview) | $0.50 | $3.00 |
| Nano Banana 2.1 | $1.50 | $7.50 |
| GPT-5 Image | $10.00 | $10.00 |
| Nano Banana Pro (Gemini 3 Pro Image) | $2.00 | $12.00 |
| Nano Banana Pro (Gemini 3 Pro Image Preview) | $2.00 | $12.00 |
| GPT-5.4 Image 2 | $8.00 | $15.00 |
相关页面
Explore multimodal capabilities, vision models, and head-to-head model comparisons.
Top AI image generators include DALL-E 3 (best text integration), Flux (best photorealism), Stable Diffusion (most customizable), and Midjourney (best artistic quality). Each excels at different styles and use cases.
Prices range from free (self-hosted Stable Diffusion) to $0.04-0.12 per image (DALL-E 3, Flux API). Most providers offer both pay-per-image and subscription pricing models.
Generally yes, but check each provider's terms. DALL-E 3, Stable Diffusion, and Flux allow commercial use of generated images. Midjourney allows commercial use for paid subscribers.