LTX-2.3 22B Distilled

About

LTX-2.3 22B Distilled is the Supernet’s default video engine and its best price-performance by a wide margin: a five-second 720p clip costs about seven cents and renders in two to three minutes — with native audio and dialogue generated in the same pass.

The distilled eight-step schedule covers five workflows: text-to-video, image-to-video, audio-driven and image-plus-audio generation, and video-to-video with ControlNet (canny, pose, depth, detailer). For longer coherent shots at higher step counts, step up to the Dev tier.

Workflows

LTX-2.3 22B Distilled exposes these workflow variants — pass the matching model id.

Workflow	Model id
Text → video	ltx23-22b-fp8_t2v_distilled
Image → video	ltx23-22b-fp8_i2v_distilled
Audio → video	ltx23-22b-fp8_a2v_distilled
Image + audio → video	ltx23-22b-fp8_ia2v_distilled
Video → video (ControlNet)	ltx23-22b-fp8_v2v_distilled

Pricing

On the Sogni API you pay per clip in Spark (1 Spark = $0.005). Prefer a flat rate? The Sogni app runs LTX-2.3 22B Distilled and 100+ other models credit-free under fair use.

Configuration	Spark	USD
5 s · 1280 × 720 · 8 steps	13.8 Spark	$0.069
10 s · 1280 × 720 · 8 steps	27.6 Spark	$0.14
5 s · 1920 × 1088 · 8 steps (scales with pixels)	31.4 Spark	$0.16

API

One Sogni API key reaches every model on the Supernet — call LTX-2.3 22B Distilled by id.

import { SogniClient } from '@sogni-ai/sogni-client';

const client = await SogniClient.createInstance({
  appId: crypto.randomUUID(),
  apiKey: process.env.SOGNI_API_KEY,
  network: 'fast',
});

const project = await client.projects.create({
  type: 'video',
  modelId: 'ltx23-22b-fp8_t2v_distilled',
  positivePrompt: 'a slothicorn surfing a wave of liquid paint, slow push-in, cinematic',
  numberOfMedia: 1,
  duration: 5,
});

const [url] = await project.waitForCompletion();
console.log(url); // result link — download within 24h

curl https://api.sogni.ai/v1/creative-agent/workflows \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $SOGNI_API_KEY" \
  -d '{
    "input": {
      "title": "First render",
      "steps": [{
        "id": "step1",
        "toolName": "generate_video",
        "arguments": { "prompt": "a slothicorn surfing a wave of liquid paint, slow push-in, cinematic", "videoModel": "ltx23", "duration": 5 }
      }]
    },
    "confirm_cost": true
  }'

REST workflow alias: ltx23. Pin the exact id with the SDK. Full reference at docs.sogni.ai.

Made on Sogni

See what LTX-2.3 22B Distilled makes

Real generations from Sogni. Hit Use prompt to open the app with LTX-2.3 22B Distilled selected and the full prompt preloaded.

A cinematic scene of A close-up portrait captures a person with rich, warm-toned skin densely speckled with dark brown freckles across the forehead, nose, and cheeks. Their eyes are wide and gazing slightly upward, framed by well-defined, arched eyebrows and long, curled lashes; the irises are deep brown with subtle golden flecks catching the light. Glossy, nude-pink lips are parted just enough to reveal a hint of teeth, coated in a high-shine balm that reflects the soft, diffused studio lighting. The skin’s surface gleams with a dewy sheen, particularly on the bridge of the nose and cheekbones, suggesting a hydrating or oil-based finish under warm, directional illumination from above and slightly to the right. The background is a smooth, matte taupe wall, providing neutral contrast that emphasizes the texture and luminosity of the subject’s complexion. The subject says "Hey fam, check out the new LTX 2 dot 3 update to Sog Knee video. Image to video in Sogni never looked so good! I mean damn." with a confident, sassy tone, her lips moving with emphasis. She casually turns her head slightly, looking directly at the camera, then breaks into a warm, genuine smile at the end. The movement is subtle and natural, with soft, lush human emotion radiating from her face. The camera performs a slow, subtle push-in. The footage remains smooth and stabilised throughout.

A cinematic scene of A woman with shoulder-length brown hair stands in profile on a dark, jagged rock outcrop, gazing toward the left at a turbulent sea. She wears a long, dark gray hooded coat that reaches mid-calf, paired with black leggings and sturdy black boots; her posture is still and upright, hands loosely at her sides, expression contemplative and distant. The scene is framed by towering, shadowed cliffs on either side, their surfaces rough and textured in deep charcoal tones, while the ocean below churns with layered waves capped in frothy white foam under an overcast, pale blue-gray sky. Diffuse, even lighting casts no strong shadows, emphasizing the muted color palette of grays, blues, and blacks, with subtle textural detail visible in the fabric of her coat and the uneven surface of the rocks. A moody, atmospheric scene. The woman stands still, peering out at the stormy ocean for a moment. She then slowly turns her head to look back towards the viewer with an intense expression. She removes her long grey coat, lets it fall to the ground, and then leaps off the cliff. The camera remains stationary as the wind and waves continue to move through the empty scene. The sound of crashing waves, strong wind, and a dramatic, somber soundtrack fills the air.

A cinematic scene of A close-up of a squirrel with dense, textured brown fur and oversized, luminous eyes that glow with swirling nebulae in reds, purples, and blues. The squirrel holds a glowing mushroom cap with both paws — its surface shimmers with iridescent cosmic patterns, dripping translucent, star-speckled droplets from its rim. Its mouth is slightly open, teeth visible, as if mid-bite or about to consume the mushroom. The background dissolves into soft-focus bokeh of warm amber and cool violet lights, suggesting an enchanted forest at twilight, illuminated by ambient bioluminescence and scattered particles of light floating through the air. The entire scene glows with ethereal, otherworldly radiance, blending organic textures with celestial imagery. The chipmunk looks up from the magical mushroom and stares awkwardly at the viewer with wide eyes. He says in a high-pitched chipmunk voice: "Wow dude, check out the new LTX 2 dot 3 update to Sog Knee video. Image to video in Sog Knee never looked so... so... what was I saying? I'm freaking out man, freakin out!" Then he bolts out of the frame quickly. The scene has a psychedelic, colorful background. The chipmunk has natural, lifelike movements. The audio includes the chipmunk's voice and ambient magical sounds.

Extreme close-up portrait of a brutal orc warrior, filling the frame, staring directly into camera, savage intimidating expression, massive tusks, broken dirty yellow teeth, rough gray-green skin with deep cracks and wrinkles, thick coarse eyebrows, leathery skin texture, blood smeared across one side of the face, fresh wounds and torn flesh around the eye socket, one glowing red ember eye, the other eye natural amber-brown, mud, grime, scars, battle damage, wet skin, realistic pores, facial hair around chin and jaw, monstrous fantasy creature, ultra-detailed cinematic realism, dramatic moody forest lighting, muted green background bokeh, shallow depth of field, sharp focus on eyes and skin texture, high contrast, gritty dark fantasy, photoreal, volumetric natural light, 85mm lens, macro portrait feel, terrifying warrior, asymmetrical composition, desaturated earthy tones, extremely detailed skin, film still, masterpie. He says: "Hey brother, check out the new LTX 2 dot 3 update to Sog Knee video. Image to video in Sog Knee never looked so skull crushingly good!"

A close shot opens on trembling elderly hands tying a gold string onto the shoulder of a life-sized ballerina puppet in a narrow workshop lit by moonlight and candlelight. Dust floats around hanging masks, velvet curtains, and antique tools as the camera slowly pulls back to reveal the old puppet maker with silver hair, round glasses, and a worn dark vest. When he finishes the knot, the ballerina puppet lifts her head and steps onto the wooden floor while the camera makes a gentle circular move around her. The old man freezes as she begins a delicate first turn beneath a shaft of moonlight. Soft creaking wood, faint ticking mechanisms, quiet music box melody.

An extreme wide shot opens at sunrise over endless white sand dunes under a pale gold sky as wind pushes dust across the surface. At the center of the desert, a colossal buried cathedral rises slowly from the sand while arches and statues emerge and spill sand in heavy streams. A lone traveler in beige wind-blown robes rides a hovering bike toward it as the camera sweeps low across the dunes and circles upward to reveal the full scale of the structure. Colored light pours through giant stained-glass windows as the traveler stops at the entrance and looks up. Ambient desert wind, deep stone grinding, soft metal resonance, no dialogue

FAQ

LTX-2.3 22B Distilled on Sogni

What is LTX-2.3?

LTX-2.3 is an open-source AI video model from Lightricks. It's notable for generating video with native, synchronized audio — speech, ambience, and effects rendered together with the picture rather than added afterward — and produces high-resolution clips from a text prompt or a starting image. On Sogni it runs on a decentralized GPU network, so there's nothing to install.

What resolution, frame rate, and duration does LTX-2.3 support?

On Sogni, LTX-2.3 supports 640px to 3840px per side, 25 to 505 frames, and flexible 1-60 fps direct generation. In practical production terms that means up to 4K output, 1080p clips up to roughly 20 seconds at 24/25 fps, and high-frame-rate cinematic options such as 48/50 fps for shorter clips.

Is it really unlimited?

Yes — covered jobs are credit-free under fair-use. There are no per-clip credits to buy, but active-job concurrency, generate queue limits, and fair-use controls apply. Covered jobs run on Sogni's shared GPU network and may queue during peak demand; heavy-usage based throttles reset each UTC day. Video resolution and clip length scale with your plan tier.

How much does it cost?

Sogni Unlimited is $20/month (or $199/year). Unlimited Pro is $50/month (or $498/year) and adds higher throughput, higher queue priority, and larger supported outputs where available. Every plan starts with a 3-day free trial and you can cancel anytime.

Can I use long, detailed prompts?

Absolutely — LTX-2.3 follows long, structured prompts, including spoken lines and audio direction. The examples above show complete, unedited prompts alongside the playable clip each produced. Hit "Use prompt" to load the full prompt into the app and reproduce or remix it.

Does LTX-2.3 do image-to-video and control LoRAs?

Yes. LTX-2.3 runs both text-to-video and image-to-video — animate a still or generate from a written scene, with native audio either way. It also supports control LoRAs for directable shots: Pose Control, Canny (edge), a Detailer, and Inpaint / Outpaint, so you can guide motion, structure, and framing instead of leaving them to chance.

Start with LTX-2.3 22B Distilled today

Create in the app, or build with the API. Your call.

Generate a video → Browse all models

About

Workflows

Pricing

API

See what LTX-2.3 22B Distilled makes

Unlimited, not metered

No credits, ever

100+ models

Per-render API

Powered by people

LTX-2.3 22B Distilled on Sogni

Start with LTX-2.3 22B Distilled today

About

Workflows

Pricing

API

See what LTX-2.3 22B Distilled makes

Unlimited, not metered

No credits, ever

100+ models

Per-render API

Powered by people

Explore the rest of the Supernet

LTX-2.3 22B Distilled on Sogni

Start with LTX-2.3 22B Distilled today