LTX Video 2.3 Fast

• 🆕 Newest fast text-to-video model in Lightricks' open-source LTX-2 family.
• ⚡ Speed-optimized tier built for iterative, short-form video workflows.
• 🧠 Diffusion-transformer foundation model generating video with audio.
• 🔊 Native synchronized audio generated alongside the visuals.
• 🔧 Open weights with LoRA fine-tuning and local execution support.
• 💬 Generates video directly from text prompts.
• 🎯 Trades peak fidelity for speed; pairs with a Full Quality sibling.
• 🏢 Released by Lightricks, makers of Facetune and LTX-Video.

Lightricks

Lightricks is a creative technology company founded in January 2013, widely recognized for its popular selfie-editing app Facetune. Originally focused on mobile video and image editing tools, the company has expanded significantly into generative AI,…

LTX Video 2.3 Fast is the speed-optimized, text-to-video member of Lightricks' LTX-2.3 release, the newest generation of the company's open-source LTX-2 family. Like the rest of the lineup, it is a diffusion-transformer foundation model that produces video with synchronized audio, and it ships with open weights for local execution. The Fast tier is framed for trying ideas quickly — iteration and short-form clips — while the LTX Video 2.3 Full Quality variant targets shots that need maximum detail.

As a same-family update, the 2.3 generation succeeds LTX Video 2.0 Fast within Lightricks' fast text-to-video line. Both are positioned as the rapid-iteration counterpart to the Full Quality checkpoints, generating from text prompts rather than reference images.

The broader LTX-2.3 family supports both text-to-video and image-to-video workflows and is distributed through Lightricks' official LTX repositories, with LoRA fine-tuning available for custom characters, styles, and control. Lightricks, founded in 2013 and known for apps such as Facetune, develops the model as part of its open generative-AI video stack, offering Fast and Full Quality tiers across both text and image conditioning so creators can choose between iteration speed and output fidelity.