LTX-2.5
Video modelOpen weightsCard summarizedLTX-2.5 improves quality, continuity, control, and efficiency through native multi-shot, stronger prompt adherence, and better local performance.
Source: LTX-2.5 Model Card
Where to call it
- LTX API
ltx-2-5-pro - Self-host (weights)
Lightricks/LTX-2.5
Good for
- Synchronized video and audio in one pass
- Multi-shot scenes with continuity across cuts
- Local execution and fine-tuning (LoRA, IC-LoRA)
Watch out for
Overview
LTX-2.5
LTX-2.5 improves quality, continuity, control, and efficiency through native multi-shot, stronger prompt adherence, and better local performance.
- Max resolution
- 4K
- Max length
- 20 s
- Modes
- text-to-video, image-to-video
- Released
- 11 Aug 2026
LTX-2.5 is Lightricks' open-weights audio-video model, a single DiT that generates synchronized video and audio from text, image and video inputs, built for local execution and fine-tuning. Over LTX-2.3 it adds a diffusion video decoder, native multi-shot scenes, a Gemma 4 12B text encoder, automatic clip duration, native EXR/HDR workflows and an improved distilled model, trained on a larger, more filtered dataset with RL post-training. Model card; open-source docs overview; model page; API changelog 11 Aug 2026
ltx-2-5-pro- Typical price
- $0.17 / sec
- Max resolution
- 4K (3840×2160 / 2160×3840)
- Max length
- 20 s (Fast at 720p/1080p, 24/25 fps); 10 s otherwise
- Aspect ratios
- 16:9, 9:16, 1:1, 3:2, 2:3
- Frame rate
- 24, 25, 48, 50
- Released
- 11 Aug 2026
- text-to-video
- image-to-video
- first and last frame
- audio-to-video
- native multi-shot
- automatic duration
- video-to-video with IC-LoRAs (open weights)
- SDR to HDR and native EXR (open weights)