Open-source video generation models and world simulators
AI video generator using MiniMax H3 Max for text-to-video and image-to-video creation
Stage
Not given
Not given
Founded
Not given
Not given
Based in
US
Not given
Pricing model
Freemium
Freemium
Pricing
Free plan: $0/month with 250 lifetime credits. Lite plan: 1,200 credits per month. Standard plan: 5,000 credits per month. Mochi video generation costs 100 credits; replay video costs 50 credits.
Free account with initial credits; annual plans from $149/year (Basic: 14,400 credits) to $999/year (Max: 180,000 credits). Generation billed per second: H3 Max at 480P from 10 credits/second, 768P from 16 credits/second, 1080P from 32 credits/second; standard H3 at 480P 10, 768P 12, 2K 26, 4K 32 cr
Key features
Text-to-video generation from written descriptions
Local execution and customization via open-source repository
ComfyUI integration for tailored experience
Interactive playground for testing capabilities
Commercial usage rights on paid plans
MP4 video format with H.264 encoding
Videos up to 5.4 seconds at 30fps
Video download for 30 days after generation
Watermark removal on paid plans
High queue priority for paid users
Text-to-video and image-to-video generation
5-15 second video output duration
Multiple resolution options up to 4K
Optional beginning and ending frames for composition control
Synchronized audio generation with video
Prompt expansion controls for detail levels
Character consistency across multiple shots
Multiple aspect ratios including 21:9, 16:9, 4:3, 1:1, 3:4, 9:16
Directed camera movement following descriptions
Per-second credit billing with upfront cost display
What makes it different
Apache 2.0 licensed open-source model for individual and commercial use
World simulator technology for embodied AI breakthroughs
Team with 55,000+ citations across foundational AI research
Run locally without vendor lock-in
State-of-the-art open text-to-video model
Independent operation with transparent credit-based billing and account-owned storage
Synchronized native audio generation in the same pass as video
Strong prompt adherence and fast iteration with H3 Max model
Failed renders automatically refunded in full credits