Loremotion launches free LTX 2.3 video generator on self-hosted GPU cluster
Loremotion.com offers unlimited AI video generation using LTX 2.3 and Wan 2.1 on a mix of owned and rented GPUs, delivering 720p 10-second clips in 50–110 seconds with no paywalls or credit limits.
A developer launched Loremotion.com this week, a free AI video generation platform offering unlimited text-to-video and image-to-video renders without subscriptions or credit systems. The service runs primarily on LTX 2.3—specifically the 1.1 distilled checkpoint—alongside Wan 2.1, both open-weight models hosted on a self-built GPU cluster.
The platform caps output at 720p resolution and 10-second duration. Generation time ranges from 50 to 110 seconds per clip depending on which GPU in the pool picks up the job. The developer uses four personal GPUs (RTX 30 and 40 series cards) supplemented by spot-rented A100s and RTX Pro units, keeping daily operating costs around $8. Revenue comes from on-site advertising rather than paywalls.
Infrastructure and workflow
User clips persist in a personal dashboard for 48 hours before automatic deletion. A public Discover gallery lets creators opt in to share finished work. When the queue backs up, users can register for email alerts that fire once their render completes. The developer notes that the LTX 2.3 distilled variant delivers the best speed-to-quality ratio for the target render time, though the platform may expand to other models as the hardware scales.
Video-to-video capability is the next planned feature. The developer is soliciting feedback on faster inference techniques and alternative open-weight video models that could reduce the sub-two-minute render window further.