Image Models
google/nano-banana
Based on Gemini 2.5 Flash Image — fast and low-cost.google/nano-banana-pro
Based on Gemini 3 Pro Image — supports up to 4K output.google/nano-banana-2
Based on Gemini 3.1 Flash Image — the most fully featured model in the Flash series, supporting 512p to 4K.openai/gpt-image-2
Based on OpenAI gpt-image-2 — supports 1K / 2K / 4K, with three quality tiers (low / medium / high) at each resolution. Quality is orthogonal to resolution.Video Models
wan3.0-video
Defaults to 720P and 5 seconds; output duration is 2-30 seconds. Reference video duration adds no charge, and audio generation does not change the price. Billing uses requested output duration, without a true-up against the actual output duration.
See Wan 3.0 Video.
kling/kling-v3
kling/kling-v3-omni
kling/kling-v3-motion-control
doubao/seedance-2.0
Without a reference video
With a reference video
Consumption depends on the total length of the reference videos and the output duration. Typical ranges:
Common scenario examples:
doubao/seedance-2.0-fast
Without a reference video
With a reference video
Consumption depends on the total length of the reference videos and the output duration. Typical ranges:
Common scenario examples: