Models & Parameters Overview

Browse currently public image and video models, then open single-model docs for exact parameters.

As more public models are added, keeping every model and parameter on one page becomes hard to read and even harder to maintain.

The current public scope includes image and video models. The overview below is based on models with active channels and usable routes.

Recommended Reading Order

Start on this page to see which public models exist.
Then open Image Models or Video Models to review shared fields and model differences.
Finally open the individual model page to see exactly how that model should be called.

Section	What Lives Here	Read First	Read Next
`Image Models`	All image generation and editing models	Shared fields and differences	Single-model page
`Video Models`	All video generation models	Shared fields and differences	Single-model page

Model	Type	Region	Best For	Entry
`gpt-image-2`	Image	Global / CN	General image generation, posters, covers, editing	View
`gpt-image-2-official`	Image	Global	Official GPT Image 2 with quality tiers	View
`gemini-3.1-flash-image-preview`	Image	Global / CN	Fast image generation and editing	View
`gemini-3-pro-image-preview`	Image	Global / CN	High-quality image generation and editing	View
`nano-banana-2`	Image	Global / CN	Faster image generation	View
`nano-banana-pro`	Image	Global / CN	Higher-quality image generation and editing	View
`seedance-2`	Video	Global / CN	Standard video generation with video input	View
`seedance-2-fast`	Video	Global / CN	Fast video generation with video input	View
`seedance-2-mini`	Video	Global / CN	Lightweight video generation with video input	View
`seedance-2-5`	Video	Global / CN	Seedance 2.5 video generation	View
`kling-3.0-turbo`	Video	Global / CN	Fast Kling video generation	View
`kling-v2-6`	Video	Global / CN	Kling standard/pro tiers	View
`kling-v3`	Video	Global / CN	Kling V3 standard, pro, and 4K	View
`kling-v3-omni`	Video	Global / CN	Kling Omni with reference video URLs	View
`kling-video-o1`	Video	Global / CN	Kling O1 with reference video URLs	View
`gemini_omni_flash`	Video	Global / CN	10-second Gemini Omni video	View

When a new model is added, update the family-level difference table first, then add the single-model page.
Do not create one new create API per provider; keep the interface layer unified.
When a model only adds one or two special fields, document the differences instead of duplicating a long page template.
If backend model schema becomes stable enough, this layer should gradually move toward generation rather than full manual editing.