kwaivgi/kling-avatar-v2 ❓🖼️📝 → 🖼️

⭐ Official ▶️ 38.1K runs 📅 Feb 2026 ⚙️ Cog 0.18.0

image-to-video-with-audio lipsync

Performance

226.7sTypical run time

38.1KTotal runs

About

Create avatar videos with realistic humans, animals, cartoons, or stylized characters

Example Output

Prompt:

"a beauty blogger talking"

Output

Performance Metrics

226.74s Prediction Time

226.76s Total Time

All Input Parameters

{
  "mode": "pro",
  "audio": "https://replicate.delivery/pbxt/OWfsqmZmKvfqivmp36eEJFtX5LpezzlTlYJOhJDdQ1um2IvG/everyday.mp3",
  "image": "https://replicate.delivery/pbxt/OWfsqnU5gDzyItM8om0stRHdPEbhClt3a3VFZdKj7qH26tq1/replicate-prediction-w7jar6nja5rmy0cw4k99cc02pr.webp",
  "prompt": "a beauty blogger talking"
}

Input Parameters

mode Default: std: Video generation mode.
audio (required) Type: string: Audio file for avatar. Supports .mp3/.wav/.m4a/.aac, max 5MB.
image (required) Type: string: Avatar reference image. Supports .jpg/.jpeg/.png, max 10MB, dimensions >= 300px, aspect ratio 1:2.5 to 2.5:1.
prompt Type: string: Positive text prompt to define avatar actions, emotions, and camera movements.

Output Schema

Output

Type: string • Format: uri

Example Execution Logs

Uploading audio to GCP...
Using model: kling-avatar in pro mode
Generating avatar video...
Generated video in 222.5sec
Generated video ID: 847752292794908751
Downloading 7630534 bytes
Downloaded 7.28MB in 3.03sec

Version Details

Version ID: b8c884263fcc40098f7c3e3bf6cb7583adf1ff0396ad5540cd27439218e78d8c
Version Created: April 17, 2026

Run on Replicate →