How to beginner · 3 min read

How to run video models on Replicate

Q: How to run video models on Replicate

Use the replicate Python package to run video models by calling replicate.run() with the model name and video input parameters. Provide the video file path or URL in the input dictionary, then process the output which may include video URLs or metadata depending on the model.

Quick answer

Use the replicate Python package to run video models by calling replicate.run() with the model name and video input parameters. Provide the video file path or URL in the input dictionary, then process the output which may include video URLs or metadata depending on the model.

PREREQUISITES

Python 3.8+
pip install replicate
Replicate API token set as environment variable REPLICATE_API_TOKEN

Setup

Install the replicate Python package and set your Replicate API token as an environment variable for authentication.

bash

pip install replicate

export REPLICATE_API_TOKEN="your_token_here"  # Linux/macOS
setx REPLICATE_API_TOKEN "your_token_here"  # Windows PowerShell

output

Collecting replicate
  Downloading replicate-0.10.0-py3-none-any.whl (20 kB)
Installing collected packages: replicate
Successfully installed replicate-0.10.0

# No output for environment variable set command

Step by step

This example shows how to run a video model on Replicate by uploading a local video file and receiving the processed output URL.

python

import os
import replicate

# Initialize client with API token from environment
client = replicate.Client(api_token=os.environ["REPLICATE_API_TOKEN"])

# Example video model on Replicate (replace with actual model name)
model_name = "stability-ai/stable-video-diffusion"

# Path to local video file
video_path = "input_video.mp4"

# Run the model with video input
output = client.run(
    model_name,
    input={"video": open(video_path, "rb")}
)

print("Output:", output)

output

Output: https://replicate.delivery/pbxt/abcd1234efgh5678/output_video.mp4

Common variations

Use video URLs instead of local files by passing {"video": "https://example.com/video.mp4"} as input.
Run asynchronously with asyncio and replicate.async_run() for non-blocking calls.
Switch models by changing the model_name to other video models available on Replicate.

python

import asyncio
import os
import replicate

async def run_video_model():
    client = replicate.Client(api_token=os.environ["REPLICATE_API_TOKEN"])
    model_name = "stability-ai/stable-video-diffusion"
    output = await client.async_run(
        model_name,
        input={"video": "https://example.com/sample.mp4"}
    )
    print("Async output:", output)

asyncio.run(run_video_model())

output

Async output: https://replicate.delivery/pbxt/xyz9876output.mp4

Troubleshooting

If you get authentication errors, verify your REPLICATE_API_TOKEN environment variable is set correctly.
For large video files, ensure your network connection is stable and consider using URLs instead of uploads.
If the model fails, check the model documentation on Replicate for required input formats and supported video codecs.

✅

Key Takeaways

Use the official replicate Python package with your API token for authentication.
Pass video inputs as file objects or URLs in the input parameter to client.run().
Async calls with client.async_run() enable non-blocking video model execution.
Check model-specific input requirements on Replicate to avoid format or codec errors.

Verified 2026-04 · stability-ai/stable-video-diffusion

Verify ↗