Built for Production
A stronger foundation from first prompt to final grade
LTX-2.5 improves the parts that determine whether generated footage is usable: motion, continuity, control, resolution and compute efficiency.
Cleaner, Natural Motion
A new diffusion video decoder reduces motion artifacts and keeps movement smoother, more natural and more stable from frame to frame.
Native Multi-Shot Continuity
Create connected shots that preserve character, environment, lighting and voice across cuts for more coherent visual storytelling.
Stronger Prompt Adherence
Give complex creative direction in shorter, simpler prompts while keeping tighter control over subjects, actions, style and camera intent.
Automatic Clip Duration
LTX-2.5 selects a fitting duration for the requested action, helping movement finish naturally with less manual setup.
Native 4K HDR and RAW
Generate high-resolution HDR footage and work in professional color pipelines with more latitude for grading, finishing and delivery.
Open and More Efficient
Run open weights on your hardware, iterate with the improved distilled model or fine-tune the pretrained foundation on your own data.
LTX-2.5 Prompts for Connected, Cinematic Video
Ready-to-use prompts for cleaner motion, multi-shot continuity, dialogue and camera direction. Copy, adapt and create.
Parisian Cafe Mood Shot
A woman stirs coffee by a rainy Parisian cafe window, soft focus background with warm interior lighting.
A woman in her 30s sits by the window of a small Parisian café. Rain runs down the glass behind her. Warm tungsten interior lighting. She slowly stirs her coffee while glancing at her phone. Background softly out of focus. The ambient sound of rain pattering against the glass and soft cafe chatter fills the air.
Hair & Backlight Close-up
Extreme close-up of wind moving through curly hair with soft afternoon backlight.
Close-up of wind moving through fine, curly hair. Individual strands visible. Soft afternoon backlight catching edge detail. The person slowly turns their head toward the camera.
Blocking & Micro-actions
Two people facing each other on a sidewalk, micro-actions showing mood changes.
Two people stand facing each other on a quiet suburban sidewalk. The taller man stands on the left, hands in pockets. The woman stands on the right, holding a bicycle. Houses blurred in the background. The man shifts his weight slightly as he speaks.
Create More with LTX-2.5
Practical guides for prompting, multi-shot planning, image-to-video and production workflows.
Choose Your Monthly Video Credits
All plans include every available LTX model. Save up to 33% with yearly billing.
Basic
For occasional projects and exploring ideas.
- Pay $238.8 yearly
- Save $120 per year
- 100 credits every month
- LTX-2.5 Fast generates about 20 videos
- LTX-2.3 Fast generates about 12 videos
- LTX-2.3 Pro generates about 9 videos
- Text-to-video and image-to-video
- Synchronized audio and resolutions up to 4K
Standard
For regular creation and iteration.
- Pay $478.8 yearly
- Save $240 per year
- 220 credits every month
- LTX-2.5 Fast generates about 44 videos
- LTX-2.3 Fast generates about 27 videos
- LTX-2.3 Pro generates about 20 videos
- Text-to-video and image-to-video
- Synchronized audio and resolutions up to 4K
Pro
For higher-volume video creation.
- Pay $1,078.8 yearly
- Save $480 per year
- 500 credits every month
- LTX-2.5 Fast generates about 100 videos
- LTX-2.3 Fast generates about 62 videos
- LTX-2.3 Pro generates about 45 videos
- Text-to-video and image-to-video
- Synchronized audio and resolutions up to 4K
LTX-2.5 Frequently Asked Questions
Clear answers about multi-shot generation, 4K HDR, local deployment and everyday creation
What is LTX-2.5?
LTX-2.5 is the latest open-weights foundation model from LTX. It is designed for video generation workflows that teams can run, fine-tune and deploy on their own infrastructure.
What is new in LTX-2.5?
LTX-2.5 adds cleaner motion, native multi-shot continuity, stronger prompt adherence, automatic duration, native 4K HDR and RAW workflows, plus a more efficient distilled model.
How should I write LTX-2.5 prompts?
Describe the subject, action, setting, camera and sound in clear natural language. For multiple shots, name each transition and restate the framing while keeping character and environment details consistent.
What is Diffusion Fidelity Rendering?
It is LTX-2.5's rendering technology for allocating compute according to scene complexity, helping fine detail and pixel quality remain stable from frame to frame.
Can LTX-2.5 create connected multi-shot scenes?
Yes. Native multi-shot generation is designed to preserve character, environment, lighting and voice across cuts, so a sequence feels connected instead of assembled from unrelated clips.
Does LTX-2.5 support 4K HDR and RAW output?
Yes. LTX-2.5 supports native 4K HDR footage and RAW-oriented workflows for professional color grading and finishing without compromising the master.
Can I use an image as the starting point?
Yes. Use a reference frame to guide visual identity and composition, then focus your prompt on the motion, camera behavior and environmental changes you want to see.
Can I run or fine-tune LTX-2.5 locally?
Yes. The open weights can be self-hosted, and the pretrained base checkpoint is built for fine-tuning on your own data. Official configurations list support starting at 16GB VRAM.
How does automatic duration work?
LTX-2.5 predicts a clip length that fits the requested action. This reduces manual duration tuning and helps motion complete naturally within the generated shot.
When should I use Fast or Pro?
Use Fast for prompt exploration, storyboards and rapid iteration. Choose Pro when final detail, cleaner motion and commercial-ready output matter most.
Your Next Shot Starts Here
Create with LTX-2.5 using prompts or reference frames, then refine the result with precise generation controls.