Loremotion offers free AI videos using open-source LTX 2.3 on personal GPUs
No credits, no subscriptions—just a dev's own GPU cluster running LTX 2.3 and Wan 2.1.
A solo developer has launched loremotion.com, a free AI video generation platform that bypasses paid APIs by running open-source models on a hybrid GPU cluster. The site uses LTX 2.3 (specifically the 1.1 distilled version) and Wan 2.1 to deliver Text-to-Video and Image-to-Video capabilities at 720p resolution with 10-second clips. The hardware setup includes four personal GPUs (RTX 30 and 40 series) supplemented with spot-rented A100s and RTX Pros, keeping daily operational costs around $8. Generation times range from 50 to 110 seconds depending on GPU allocation.
To sustain free unlimited generations, the site relies on advertisements rather than subscriptions or credit systems. Users can access a dashboard where clips are stored for 48 hours, optionally push their best results to a public Discover gallery, and receive email alerts when queue times are long. The developer is already working on Video-to-Video features actively seeks community suggestions for improving generation speed and exploring better models.
- Uses open-source LTX 2.3 (1.1 distilled) and Wan 2.1 for Text-to-Video and Image-to-Video at 720p/10s
- Runs on a mix of personal RTX 30/40 series GPUs and rented A100/RTX Pros, costing ~$8/day
- Free to use with no credits or limits; monetized solely via ads
Why It Matters
Democratizes AI video creation by offering free, unlimited generation without API costs or subscriptions.