The models behind every generation
One prompt behaves differently on every model. These are the families we run, what each is actually for, and where each one falls down — with the capability tables and prices generated from the live catalogue rather than written by hand.
Model families
Capability tables and prices are read from the live model catalogue. Illustrations are art direction, not model output.
Video models
Seedance
ByteDanceLong single takes with sound, and references you address in the prompt
- Start from text
- Start from an image
Models · 4 Tools · 2
- 4k
- 30s
- Audio
Kling
KuaishouKuaishou’s video line — from silent drafts to 4K, with sound where you want it
- Start from text
- Start from an image
Models · 5 Tools · 2
- 4k
- 15s
- Audio
MiniMax Hailuo
MiniMaxMiniMax’s video line — silent Hailuo clips and H3 with sound
- Start from text
- Start from an image
Models · 5 Tools · 2
- 2K
- 15s
- Audio
WAN
Alibaba · Tongyi LabLong takes from a prompt, a first frame or references for picture and sound
- Start from text
- Start from an image
Models · 4 Tools · 3
- 2K
- 30s
- Audio
Gemini Omni
GoogleGoogle’s video model, from a 360p draft to 4K
- Start from text
- Start from an image
Models · 1 Tools · 2
- 4k
- 10s
- Audio
Grok Imagine
xAIxAI’s video line — short clips with sound from a prompt
- Start from an image
Models · 2 Tools · 1
- 10s
- Audio
FLUX Video
Black Forest LabsBlack Forest Labs’ video model, with sound and many references
- Start from text
- Start from an image
Models · 1 Tools · 2
- 1080p
- 20s
- Audio
Image models
GPT Image
OpenAIOpenAI’s image line — quality levels you choose per image
- Start from text
- Start from an image
Models · 6 Tools · 3
- 1536x1024
Nano Banana
GoogleGoogle’s image line — legible text, sizes up to 4K
- Start from text
- Start from an image
Models · 3 Tools · 3
- 4K
Seedream
ByteDanceByteDance’s image line — stills, reference-driven images and edits
- Start from text
- Start from an image
Models · 4 Tools · 4
- 2K
Qwen Image
Alibaba · QwenAlibaba’s image model for dense text and layouts
- Start from text
- Start from an image
Models · 1 Tools · 3
- 2K
Speech and music models
Seed Audio
ByteDanceByteDance’s speech model — a script read in a voice you describe or attach
- Audio
Models · 1 Tools · 1
MiniMax Speech
MiniMaxMiniMax’s speech model, in HD and Turbo
- Audio
Models · 2 Tools · 1
Lyria
Google DeepMindGoogle DeepMind’s music model — a full piece or a short clip
- Audio
Models · 2 Tools · 1
- 30s
MiniMax Music
MiniMaxComplete songs, sung, from a description and your lyrics
- Audio
Models · 1 Tools · 1
Frequently Asked Questions
By constraint, not by ranking. Text in the image → Nano Banana 2, Seedream 5 Pro or Qwen Image 3, whose makers name it as a focus. A long single take with sound → Seedance 2.5 or WAN 3.0. References for a character or a motion → Kling 3.0 Omni, Seedance or WAN. A cheap 360p draft before the final, with sound → Gemini Omni. Speech → Seed Audio or MiniMax Speech; music → Lyria or MiniMax Music.
Yes, and the difference is large. Every release page lists its price for each length, size or quality it sells, and the composer shows the price for the settings you pick before you generate — the number on screen is the number charged.
Yes. The prompt and most settings carry across when you switch models; anything the new model cannot honour is struck through with the reason rather than silently dropped.
Ready to Get Started?
Every tool shows its price in credits before you run it.