LTX-2.5 generates 10-second video from an image in 6.8 seconds on Nvidia superchips, open-weights release.
The AMW Read
Open-weights video model with fast inference updates the competitive landscape in generative media, but does not resolve a major open debate or introduce a new top-tier entrant.
LTX-2.5 generates 10-second video from an image in 6.8 seconds on Nvidia superchips, open-weights release.
LTX, the open world model company spun out of Lightricks, has released LTX-2.5, an open-weights video and world model. The model can generate a 10-second video from a single image in just 6.8 seconds on Nvidia superchips, a speed that underscores the company's focus on efficiency and accessibility in generative media.
This release is significant for the AI video market, as it combines open-weight distribution with rapid inference, potentially lowering the barrier for developers and enterprises to deploy custom video generation without relying on closed, proprietary APIs. The open-weights approach allows for fine-tuning and on-premises deployment, which is increasingly attractive to organizations concerned about data control and cost. In a space dominated by well-funded closed labs, LTX's speed and openness could shift competitive dynamics toward performance-per-dollar and transparency.
For builders, this means LTX-2.5 could serve as a practical foundation for real-time or near-real-time video workflows, such as interactive content creation, personalized marketing, or synthetic data generation for training other models. For investors, the emphasis on inference speed over sheer model scale suggests a growing niche for efficiency-first video models that compete on latency and cost rather than just output quality. Per the AI Market Watch index, LTX has logged one item in the last 90 days, so this release marks a notable update in the company's trajectory.
#OpenWeights #VideoGeneration #WorldModels #AIInfrastructure #GenerativeMedia #InferenceSpeed