VideoPrivate

Longcat Distilled

Meituan's open-source, uncensored video generation model — unified architecture for text-to-video and image-to-video with efficient long-duration output.

Maker
Meituan
Modality
Video
Max duration
30 seconds
Max resolution
720p

Overview

What is Longcat Distilled

Longcat Distilled is a distilled version of Meituan's LongCat-Video, an open-source 13.6B-parameter video generation model. It supports text-to-video and image-to-video tasks with a unified Diffusion Transformer architecture, optimized for efficient inference and long-duration clips up to 30 seconds at 720p resolution.

Running it privately on Venice

Running on Venice, Longcat Distilled generates video without storing your prompts or profiling your usage — true zero retention. As an open, uncensored model, it enables permissionless creativity across both text-to-video and image-to-video workflows, with no personal data tied to generations. You maintain full sovereignty over inputs and outputs.

Private (zero retention)No prompt trainingTEE · hardware enclaveEnd-to-end encrypted

Assessment

Strengths and limitations

Strengths
  • Unified model for multiple video tasks: text-to-video, image-to-video, and video-continuation — simplifies deployment and reduces model sprawl.
  • Efficient coarse-to-fine generation strategy enables fast 720p video output within minutes, even at longer durations.
  • Strong temporal coherence and subject stability in generated clips, ideal for long-duration animation.
  • Open-source under MIT license: fully self-hostable, modifiable, and commercially usable without restrictions.
  • Uncensored and permissionless: runs on Venice with no content filtering or user profiling.
Limitations
  • No audio generation: output is video-only, limiting use in multimedia storytelling or lip-syncing scenarios.
  • Limited to 720p resolution: lacks native 4K or higher output available in some competing models.
  • Distilled variants trade sampling steps for speed, which may reduce fine detail fidelity compared to full LongCat-Video.

Samples

Sample outputs

Generated on Venice with our standard prompt suite — the same prompts we run through every model of this type, so you can judge it like-for-like.

Urban cinematic

A slow tracking shot through a rain-soaked Tokyo street at night, neon reflecting in puddles, steam rising from a food cart, a person with a translucent umbrella, shallow depth of field, teal-and-orange grade, smooth steady camera

Compare every video model on these prompts

Capabilities

What it supports

  • Text to video
  • Image to video
  • Reference to video
  • Native audio generation

Variants

Longcat Distilled model variants

Longcat Distilled runs on Venice as 2 variants of the same underlying model. Pick by what you're starting from: a written prompt, a still image, reference images, or an existing clip. Each variant is its own model id on the API; the generation quality is the same across the family.

VariantWhat it isClip lengthsResolutionsAspect ratiosAudioModel ID
Text to VideoflagshipGenerate a clip from a written prompt5s – 30s720p16:9, 9:16, 1:1longcat-distilled-text-to-video
Image to VideoAnimate a still image into motion5s – 30s720plongcat-distilled-image-to-video

Capability data comes straight from the Venice model API and refreshes with every catalog ingest. The specs and pricing on this page are captured from the flagship variant; pass the model id of the variant you want to the API.

Longcat Distilled Text to Video

Generate a clip from a written prompt. Supports clips of 5s – 30s, 720p output, 16:9, 9:16, 1:1 aspect ratios.

longcat-distilled-text-to-video

Longcat Distilled Image to Video

Animate a still image into motion. Supports clips of 5s – 30s, 720p output.

longcat-distilled-image-to-video

Specifications

Datasheet

Maker
Meituan
Released
October 25, 2025
Architecture
Diffusion Transformer (DiT)
Parameters
13.6B
Resolutions
720p
Clip lengths
5s, 10s, 15s, 20s, 30s
Mode
text-to-video
Aspect ratios
16:9, 9:16, 1:1
Audio
No
Privacy on Venice
Private — zero retention
Available on Venice since
Dec 2025
License
MIT License

API

Call it from your code

Venice exposes this model through the REST API. Queue a generation with the model id.

curl https://api.venice.ai/api/v1/video/queue \
  -H "Authorization: Bearer $VENICE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "longcat-distilled-text-to-video",
    "prompt": "Aerial drone shot over a misty mountain valley at golden hour"
  }'

# Use the returned queue_id with https://api.venice.ai/api/v1/video/retrieve.
# Call /video/complete after downloading if needed.

Pricing

What it costs on Venice

Pay per clip on Venice — price scales with resolution and duration (5s–30s), from $0.09.

720p · 5s
$0.09
Per clip

New Venice accounts include a free daily allowance and 500 welcome credits — no credit card required.

Alternatives

How it compares

ModelMax resolutionStrongest atOpen weightsPrice (Venice)
Longcat Distilled720pLong-duration coherenceYesfrom $0.09
Wan 2.71080pHigh-res realismYesfrom $0.55
Kling O3 Pro2KCinematic qualityNofrom $0.46
Vidu Q31080pText accuracyNofrom $0.27

Open, uncensored, and efficient — ideal for developers and creators wanting full control.

Use cases

What it is good for

  1. 01Animating still images into social media clips using image-to-video.
  2. 02Generating short product demonstration videos from text prompts.
  3. 03Creating long-duration, temporally stable animations for digital storytelling.
  4. 04Developers building open, privacy-first video generation pipelines.
  5. 05Content creators needing uncensored, commercially safe video assets.

Prompting

Getting better results

For image-to-video, include clear motion cues like 'slow zoom' or 'pan left' to guide animation.

Use precise timing references (e.g., '5-second clip') to align with available durations.

Leverage aspect ratio control (16:9, 9:16, 1:1) for platform-specific content — TikTok, YouTube, Instagram.

Start with shorter clips (5s–10s) for faster iteration before scaling to 30s.

Version history

LongCat-Video
2025-10

Full 13.6B-parameter foundation model.

Longcat Distilled
2025-12

Distilled for faster inference, on Venice since Dec 2025.

FAQ

Frequently asked questions

Longcat Distilled is a distilled version of Meituan's LongCat-Video model, optimized for fast, efficient video generation. It supports both text-to-video and image-to-video tasks using a unified Diffusion Transformer architecture, with strong performance in long-duration clips up to 30 seconds.

On Venice, Longcat Distilled starts at $0.09 per clip, with pricing scaling based on resolution and duration. A 720p, 5-second clip costs $0.09, while longer or higher-resolution variants cost more.

Yes. Longcat Distilled is open-source under the permissive MIT license, with model weights publicly available. You can use, modify, and deploy it commercially without restrictions.

No, Longcat Distilled is a video generation model and does not support tool use, function calling, or external API integration. Its capabilities are limited to generating video from text or image inputs.

Longcat Distilled generates video at 720p resolution across all variants. It supports aspect ratios of 16:9, 9:16, and 1:1, making it suitable for social media, mobile, and desktop platforms.

No. Longcat Distilled produces video-only output. Audio must be added in post-processing or through third-party tools.

Yes. The variant 'longcat-distilled-image-to-video' specifically animates still images into motion. You provide a reference image and a motion prompt, and the model generates a 5s–30s clip based on it.

The maximum clip length is 30 seconds. This applies to both text-to-video and image-to-video variants, with options for 5s, 10s, 15s, 20s, and 30s durations.

Longcat Distilled excels in long-duration temporal coherence and open licensing, while Wan 2.7 offers higher 1080p resolution and stronger photorealism. Choose Longcat for open, uncensored workflows; Wan 2.7 for higher fidelity when resolution matters more than openness.

Run Longcat Distilled privately

No prompt logging. No data used for training.