Model comparison

Key differences within each capability; the billing mode decides whether a higher tier costs more

Text to Image

ModelMax secondsTiersAudioBilling
Qwen Image Max—1328*1328 / 1664*928—per image
GPT Image 2—1024x1024 / 1536x1024 / 1024x1536 / 2048x1152 / 2048x2048 / 3840x2160 / 2160x3840—per image

Image Editing

ModelMax secondsTiersAudioBilling
Qwen Image Edit Max—1328*1328 / 1664*928—per image
Nano Banana 2 Edit—1K / 2K / 4K—per image

Text to Video

ModelMax secondsTiersAudioBilling
Kling V3 Text to Video15std / pro✓per second
Wan 3.0 Video30480p / 720p / 1080p✓per second
Wan 3.0 Video Pro301080p / 2k / 4k
The tiers are close together: 2K is about 1.11× and 4K about 1.28×, nowhere near double. If you want 4K, take it — dropping to 1080p saves little
✓per second
Wan 2.5 Text to Video10832*480 / 1280*720 / 1920*1080✓per second
Sora 212720x1280 / 1280x720✓per second
Kling V1.610std / pro✗per job
Veo 3.18720p / 1080p / 4k
4K is 1.5× (the Fast model’s 4K is 2.33×, pricier despite the name — use this one for 4K)
✓per second
Veo 3.1 Fast8720p / 1080p / 4k
4K is 2.33×, pricier than the standard model’s 1.5× — the standard model works out cheaper if you actually need 4K
✓per second
Seedance Lite Text to Video60480p / 720p / 1080p—per job

Image to Video

ModelMax secondsTiersAudioBilling
Image to Video Pro15720p / 1080p✓per second
Image to Video · Keyframes8480p / 720p✗per second
Kling Omni Reference to Video15std / pro✓per second
Wan 2.6 Image to Video15720P / 1080P✓per second
Wan 2.5 Image to Video10480P / 720P / 1080P✓per second
Wan 2.2 Image to Video Flash5480P / 720P✗per second
Wan Keyframe to Video5720P✗per second
Hailuo Director Image to Video6720P / 1080P—per job
Wan 2.7 Image to Video10480P / 720P / 1080P✓per second