video model · by Lightricks

LTX-2 Fast AI Video Generator

Generate videos up to 20 seconds long at 1080p with synchronized audio using LTX-2 Fast - Lightricks' open-source DiT-based audio-video foundation model. Built for creators who want longer clips, unified sound, and the flexibility of an open model, now available inside Fliki.

Generated with LTX-2 Fast

A handful of LTX-2 Fast clips generated inside Fliki. No edits, no post.

Prompt

A bilingual creator in a yellow sweater leans into a desk mic in her bedroom studio, fairy lights blurred behind her, smiles, then says in clear Spanish, "Hola, hoy vamos a hablar de algo importante," before switching to English with a grin, "but first — coffee." Selfie phone framing, slight handheld feel, warm key light. SFX: the gentle clink of a mug, soft synthwave music underneath, room tone.

Prompt

A pair of hands carefully cracks an egg into a hot cast-iron skillet, butter foaming, the egg whites bubbling at the edges, top-down phone-tripod angle, bright morning kitchen window light, slow steady pace, no dialogue. SFX: the sharp crack of the shell, the immediate sizzle, butter popping, distant kettle starting to boil, ambient quiet morning kitchen tone.

Prompt

A young man in a hoodie standing on a busy city sidewalk turns to a friend off-camera and says, "You ever just feel like, what if I just left tomorrow?" The friend laughs out of frame, traffic blurs behind them, handheld phone shot at chest height. SFX: car horns, footsteps, a passing bus, ambient street chatter, no music.

Prompt

An older man in a wool coat sits at a wooden bar holding a glass of whiskey, the bartender wipes a glass behind him, slow dolly-in on his face. He stares at the drink, then quietly says, "She wrote me a letter. After all these years." Warm tungsten lighting, soft grain, cinematic 35mm look. SFX: ice clinking in the glass, low jazz on a tinny speaker, the door bell as someone enters, distant rain.

Prompt

An elderly fisherman walks alone along a foggy stone harbour at dawn, gulls overhead, his weathered boat rocking gently against the dock, slow tracking shot from the side. A narrator's voiceover says calmly in French, "Il pêche dans cette baie depuis quarante-trois ans." Cinematic documentary grade, soft pastel sunrise. SFX: lapping water, gull cries, distant boat horn, gentle ambient string score underneath.

Prompt

A single matte black coffee cup sits on a polished wooden table, steam rising in a slow ribbon toward the top of the frame, locked square framing, soft side window light. A male voiceover says quietly, "Made slowly. On purpose." SFX: a soft pour of coffee at the start, the gentle clink of cup on saucer, a single warm acoustic guitar chord, ambient quiet.

100M+VIDEOS CREATED
12M+USERS WORLDWIDE
80+LANGUAGES SUPPORTED

Trusted by 50,000+ companies worldwide

What makes LTX-2 Fast distinct

Unified audio-video foundation

LTX-2 is a DiT (diffusion transformer) based audio-video foundation model - it generates synchronized video and audio in a single pass, instead of generating video and then overlaying sound.

Up to 20-second generations

LTX-2 Fast can produce clips up to 10 seconds long at 2160p - longer than most AI video models. Use the runway for narrative beats, ambient atmosphere, or extended product demos.

1080p output

LTX-2 Fast renders at 1080p, delivering social-ready quality without needing an upscale pass. The output holds detail across the full 10-second duration.

Synchronized native audio

Every LTX-2 clip comes with synchronized audio - ambient sound, effects, and music that align with on-screen action. Audio is part of the model's core, not an afterthought.

Open-source transparency

LTX-2 is open source, which means the model weights and inference code are publicly available. For Fliki users that translates into a model with an active research community and continuous third-party improvements.

Text-to-video, image-to-video, multi-modal

LTX-2 Fast supports text, image, and audio inputs across text-to-video, image-to-video, and audio-conditioned generation. The open architecture allows for unusually flexible input combinations.

Community-backed model evolution

Because LTX-2 is open source, improvements from Lightricks and third-party contributors show up fast. Expect a higher rate of quality improvements over time than with closed models.

ComfyUI and local-inference compatibility

The underlying LTX-2 weights are compatible with ComfyUI and local inference for advanced users. On Fliki, you get the same model through a simple prompt interface - no local setup needed.

Apache 2.0 license

LTX-2 ships under the Apache 2.0 license — full model weights, training code, and documentation are publicly released. That license is friendly to commercial work and is rare among production-grade video models.

How it works

How to generate a video with LTX-2 Fast

LTX-2 Fast runs inside Fliki's standard generator. Here's the flow.

Fliki prompt input showing a cinematic text-to-video description for LTX-2 Fast AI video generator
Step 1

Write your prompt

LTX-2 Fast handles extended durations well, so write prompts with a beginning, middle, and end. Describe the narrative arc, not just the subject.

Fliki model selector dropdown with LTX-2 Fast chosen for AI video generation
Step 2

Select LTX-2 Fast as your model

Pick LTX-2 Fast from Fliki's model selector. Your prompt is sent to the Lightricks open-source model for generation.

Choose 16:9, 9:16, or 1:1 aspect ratio for LTX-2 Fast video generation on Fliki
Step 3

Pick your aspect ratio

Choose 16:9 for landscape, 9:16 for vertical social, or 1:1 for square feed. LTX-2 Fast composes each ratio natively.

Set video duration on the Fliki slider for LTX-2 Fast multi-shot AI video generation
Step 4

Set the duration

LTX-2 Fast supports up to 20 seconds - use the extended runway for narrative, atmosphere, and extended shot beats. Shorter durations generate faster.

Upload an optional reference image to anchor subject and style with LTX-2 Fast on Fliki
Step 5

Add a reference image or audio (optional)

Upload an image to anchor the subject or an audio clip to drive timing and mood. LTX-2's multi-modal input handling is part of what makes it unusually flexible.

Pick output resolution and hit generate to create AI video with LTX-2 Fast on Fliki
Step 6

Select resolution and generate

Choose 1080p, 1440p or 2160p and hit Generate. Your clip arrives in Fliki with synchronized audio baked in.

AI MODEL GALLERY

Built on the best AI models - ready inside Fliki

Every leading video, voice, and image model - integrated, unified, and tuned for creators. Generate with the latest AI video, AI voice, and AI image models from OpenAI, Google, Kling, Bytedance, ElevenLabs, and more - all from one place.

LTX-2 Fast FAQ

Frequently asked questions

Everything you need to know about generating with LTX-2 Fast inside Fliki.

What is LTX-2 Fast?

LTX-2 Fast is Lightricks' open-source DiT-based audio-video foundation model. It generates up to 10 seconds of 1080p video with synchronized native audio, supporting text-to-video, image-to-video, and multi-modal workflows.

Is LTX-2 Fast open source?

Yes. LTX-2 is an open-source model - weights and inference code are publicly available, with an active developer and research community. On Fliki, you get hosted access with no local setup required.

What is the LTX-2 model?

LTX-2 is Lightricks' audio-video foundation model, released in late 2025, built on a diffusion transformer (DiT) architecture that generates synchronized audio and video in a single pass.

How long can LTX-2 Fast videos be?

LTX-2 Fast supports clips up to 10 seconds and 2160p - better resolution quality than most current AI video models.

Does LTX-2 Fast generate audio?

Yes. Synchronized native audio is one of LTX-2's core features, generated as part of the same model pass as the video.

What's the difference between LTX-2 Fast and LTX-2.3?

Fast prioritizes generation speed and is the production speed tier. LTX-2.3 is the latest generation tuned for maximum fidelity, sharper motion, and final delivery. Use Fast for iteration and volume; use LTX-2.3 for hero shots and final output.

Still curious?

Try Fliki free in your browser, no credit card required.

Start free

Can LTX-2 Fast do image-to-video?

Yes. Text, image, and audio inputs are all supported. Image-to-video is strong for controlled continuation and style consistency.

Do I need ComfyUI to use LTX-2 Fast?

No. ComfyUI works for local inference on the open-source weights, but inside Fliki LTX-2 Fast runs through a standard prompt interface - no local setup or ComfyUI required.

LTX-2 Fast · Free forever plan

Generate your next video with LTX-2 Fast.

Up to 20 seconds of 1080p video with synchronized native audio. Open-source power, hosted inside Fliki.

Generate your first video free

Free forever plan · No credit card required · Cancel anytime