AI · Reference
Video Generation
Text and image to video models, from the safe Western defaults to the leaderboard leaders.
16 models · continually updated
GuideClaude + MCP for video
from a written brief to a finished shot
GuideSeedance 2.5 prompting
the official method, decoded
Google DeepMindVeo 3.1
the safest Western default
KuaishouKling 3.0
the value leader with four top-ten entries
ByteDanceSeedance 2.0 and 2.5
the newest model in the field
RunwayRunway Gen-4.5
the control surface
OpenAISora 2
the cautionary tale
Alibaba ATHHappyHorse 1.0
the surprise leader
MiniMaxHailuo 2.3
the character-motion specialist
Luma AILuma Dream Machine 2.0 and Ray3
depth and volume
LightricksLTX-2.3
the cleanest open licence
AlibabaWan 2.7
the open workhorse
TencentHunyuanVideo 1.5
open weights from a hyperscaler
PixVersePixVerse V6
multi-shot for creators
Shengshu TechnologyVidu Q3
the efficient challenger
The rest of the field
Muse Video, Pika 2.5, Grok Imagine and Marey (roundup page)