LILightricks video model

LTX-2.5 Fast. Lightricks video with sound, up to 20 seconds.

Lightricks' LTX-2.5 on fal's fast endpoint: clips with native audio from a prompt, an opening frame, or a first and end frame, in 16:9 or 9:16, from 720p drafts to 4K finals.

  • 6, 8, 10, 12, 14, 16, 18 or 20 second clips
  • 16:9 and 9:16
  • Native audio
  • First and last frame control
  • 720p, 1080p, 1440p or 4K
  • 1440p and 4K up to 10 seconds
  • Runs on fal

LTX-2.5 is a video model from Lightricks, and this is fal's fast endpoint for it. Treza runs it through fal because OpenRouter does not carry LTX. It takes a prompt, a first frame, or a first and an end frame for a transition between the two, and renders 16:9 or 9:16 with native audio you can switch off. Lengths run from 6 seconds up to 20 in steps of 2 at 720p and 1080p; at 1440p and 4K a clip stops at 10.

fal listed it on 2026-10-26 at 9 cents a second at 720p, 13 at 1080p, 19 at 1440p and 30 at 4K, and sound does not change the price. Treza renders 720p unless you pick another tier and bills exactly the length you set. It quotes the list price plus its credit markup before a run, and a failed generation costs nothing. It runs on the canvas, in the Video Generation node, and through any pipeline you publish as an API. Chat does not run it; there, ask for Seedance 2.5.

Sound on or off, same price

Native audio comes with the clip, and a switch on the node turns it off for a silent take. fal bills the same either way.

Up to 20 seconds in one take

Even lengths from 6 seconds up to 20 at 720p and 1080p, and up to 10 at 1440p and 4K. Treza always sends a length, so the quote matches the bill.

First and end frames

Open on an image, or add an end frame and the clip makes the transition between the two: a sketch that becomes a render, a before that becomes an after.

720p drafts, 4K finals

Four tiers on one node. 720p lists about 30 percent under 1080p, so draft there and render the keeper at 1080p, 1440p or 4K.

What to make with LTX-2.5 Fast

Real prompts, and what comes back. Every one runs as a node on the Treza canvas.

“A barista pours steamed milk into a latte in a sunlit cafe, slow push-in, 9:16”

A short vertical beat with cafe sound.

“Start on this product photo (attached): steam rises from the cup, 8 seconds”

Image to video that opens on your frame, with sound.

“First frame: a pencil sketch of a house. End frame: the finished render (both attached)”

A transition between two images you supply.

“Waves break over dark rocks at dusk, wide, 20 seconds, 16:9”

One long take with the sound of the sea.

“The keeper from the 720p drafts, re-rendered at 4K for a landing page, 10 seconds”

The same prompt at the top tier.

“A rainy city street at night, audio off, to sit under narration”

A silent take, ready for a voiceover.

LTX-2.5 Fast is one node on a canvas

Chain it with language, image, and video models, then publish the whole pipeline as an API. See how a full AI video pipeline fits together.

  1. Step 01

    Draft the brief

    A language model node expands a one-line idea into a detailed, on-brand prompt before it ever reaches the model.

  2. Step 02

    Generate with LTX-2.5 Fast

    Run variants in parallel, compare against other models on the same prompt, and keep every result in run history.

  3. Step 03

    Publish it as an API

    The pipeline becomes a versioned endpoint. Your product calls it with the typed /invoke API or any OpenAI SDK.

LTX-2.5 Fast as an API, priced per clip

The LTX-2.5 Fast API

Build a pipeline with LTX-2.5 Fast on a node and publish it. It becomes a versioned HTTP endpoint: the typed /invoke route enqueues a run and returns a run id to poll, and an OpenAI-compatible route works with SDKs you already use. Agents connect over MCP the same way. Your prompt engineering ships inside the pipeline, so every caller gets your quality, and swapping the model later does not change the API your product calls.

LTX-2.5 Fast pricing

Prepaid credits, no subscription. Each clip is charged at LTX-2.5 Fast's own rate from your balance, shown before you run, and a failed generation costs nothing. Credit packs start at $5 and never expire, which suits spiky production schedules. Run history records the cost of every generation, per node, so spend is auditable rather than a surprise.

LTX-2.5 Fast, answered

What is LTX-2.5 Fast?

LTX-2.5 Fast is Lightricks' LTX-2.5 video model on fal's fast endpoint. It turns a prompt, a first frame, or a first and end frame into a clip with native audio, up to 20 seconds at 720p or 1080p and up to 10 at 1440p or 4K. Treza runs it in the Video Generation node.

What does LTX-2.5 Fast cost?

fal lists 9 cents a second at 720p, 13 cents at 1080p, 19 at 1440p and 30 at 4K, with or without sound, so a 6 second 720p clip is 54 cents of provider cost. Treza bills the length you set at that rate, quotes it plus its credit markup before a run, and a failed generation costs nothing.

What is the difference between LTX-2.5 Fast and LTX-2.3 Fast?

Price and the lowest tier. LTX-2.5 Fast offers 720p at 9 cents a second, and lists 13, 19 and 30 cents at 1080p, 1440p and 4K. LTX-2.3 Fast starts at 1080p and lists 6, 12 and 24 cents at those tiers. Both add sound, take a first and end frame, render 16:9 or 9:16, and run up to 20 seconds at 1080p.

How does LTX-2.5 Fast compare with Seedance 2.5?

Seedance 2.5 renders any whole second from 4 to 30, in six aspect ratios, at 480p, 720p or 1080p, with sound, first and last frames, and reference images, and chat runs it. LTX-2.5 Fast renders even lengths up to 20 seconds in 16:9 or 9:16, reaches 4K for clips of up to 10 seconds, and runs on the canvas. In a pipeline you can run one prompt on both and compare.

Can I use LTX-2.5 Fast in chat?

It runs on the canvas, in the Video Generation node. In chat, ask for Seedance 2.5.

How much does LTX-2.5 Fast cost on Treza?

Treza uses prepaid credits with no subscription. Each clip is charged at LTX-2.5 Fast's rate from your balance, and only successful generations are charged. Credit packs start at $5 and never expire.

Can I call LTX-2.5 Fast through an API?

Yes. Build a pipeline that uses LTX-2.5 Fast and publish it as a versioned HTTP endpoint. Call the typed /invoke API from your product, or point any OpenAI SDK at the OpenAI-compatible endpoint. If a better model ships later, swap it on the node without changing the API your product calls.