The right model for your next idea.

GPT Image 2.5 Flare is the default image model and a practical starting point for everyday visuals and edits. Choose 1K for an initial draft, then try 2K or 4K if you need more detail.

Flare suits everyday generation and iteration. Sunburst focuses on intricate visual detail and may take longer. Both offer text-to-image, image editing and 1K, 2K or 4K output here.

Seedance 2 Mini is the default video model. It supports 4–15 second clips at 480p or 720p, with prompts, frames or media references. Start with a short clip to check the motion.

Yes. Choose an image model, switch to image-to-image, upload a reference and describe what to change. State what should stay the same, such as the subject, colors or camera angle.

Yes. Open a video model and choose image input or frames. Add a starting image and describe the motion, camera movement and scene changes you want.

Choose Seedance 2.5 when you need up to 30 seconds, 1080p output or more reference inputs. Mini supports up to 15 seconds and 720p, with lower per-second credit rates.

It generates images from text and edits reference images. You can use up to five images per task and choose square, portrait or landscape formats. Each generation costs 4 credits.

Flare and Sunburst accept up to 16 reference images per editing task. Use clear, relevant references and explain which subject, style or composition each one should guide.

Yes. GPT Image 2.5 Flare and Sunburst offer a transparent background at 1K. Choose 1K before selecting that option; background selection is unavailable at 2K and 4K.

They control image output resolution. Higher settings provide larger images and cost more credits. For GPT Image 2.5, 1K costs 3 credits, 2K costs 10 and 4K costs 16.

Seedance 2.5, Wan 3.0 Video and Wan 3.0 Video Prime offer 1080p in this library. MiniMax H3 offers 768P and 2K. Open a model to choose its available resolution.

Seedance 2.5 and both Wan 3.0 Video variants support up to 30 seconds. Grok Video 1.5 Preview, Seedance 2 Mini and MiniMax H3 support up to 15 seconds.

Seedance 2 Mini, Seedance 2.5, both Wan 3.0 models and MiniMax H3 offer frame inputs. Use frames to guide a clip’s opening and ending; choose reference mode separately for other media.

Seedance 2 Mini, Seedance 2.5, both Wan 3.0 models and MiniMax H3 support media-reference workflows. Available combinations and upload limits appear in each model’s workspace.

Wan 3.0 Video and Wan 3.0 Video Prime offer document and public-webpage input. Select that input mode, provide an accessible link and describe the video you want to create.

For Seedance 2 Mini, Seedance 2.5 and MiniMax H3, both input and generated video durations are billed. Seedance uses a different rate when a video reference is present. Check the estimate after your uploads finish.

The first five reference images are free; each additional image costs 4 credits. Audio input is free. Video is charged at 8 credits per second for 768P or 13 for 2K, including input video duration.

Use 9:16 for vertical clips, 16:9 for widescreen video and 1:1 for square images. Ratios vary by model; GPT Image 2.5 also has some formats that are available only at 1K.

Search its name or browse the image and video lists. Nano Banana, GPT-4o Image, Seedream V4, Flux Kontext, MindVideo and Seedance v1 Lite still have direct model-page links here.

Describe the subject, setting, style, lighting and composition. For video, add the action and camera movement. When editing a reference, clearly separate the changes you want from the details you want to preserve.