Generate Your First AI Video from the Terminal in 60 Seconds

Why generate video from a terminal?

Terminal-based video generation is not just a novelty. It unlocks workflows that browser-based tools simply cannot match. You can script batch generations, pipe outputs to other tools, integrate with CI/CD pipelines, and automate entire content pipelines without ever touching a GUI. If you can write a shell script, you can build an automated video production line. agent-media CLI gives you access to 7 AI models through one interface, one authentication, and one credit balance. There is no need to manage separate accounts across Kling, Sora, Seedance, and others.

Prerequisites

You need two things before starting:

• 1.Node.js 18 or higher. Check your version with node --version. If you need to install it, grab it from nodejs.org or use a version manager like nvm.
• 2.An agent-media account. Sign up at agent-media.ai and choose a plan. The Starter plan at $19/month gives you 1,000 credits, enough for several video generations to explore every model.

Step 1: Install the CLI

Install the agent-media CLI globally with npm. This gives you the agent-media command available from any directory on your system.

After installation completes, verify it worked by checking the version:

Step 2: Authenticate your account

The CLI needs to link to your agent-media account. Run the login command and a browser window will open automatically. Sign in with your account credentials and the CLI will store a secure device key locally so you do not need to log in again.

You can verify your authentication and check your credit balance at any time:

Step 3: Generate your first video

Now for the fun part. Use the generate command with a model slug and a prompt. The --sync flag tells the CLI to wait for the generation to complete and print the output URL directly in your terminal, instead of returning immediately with a job ID.

That is it. You just generated a 5-second AI video from your terminal. The output URL is a direct link to your MP4 file. Open it in a browser, download it with curl, or pipe it into your video editing workflow.

Step 4: Understanding sync vs async mode

The --sync flag is convenient for quick one-off generations because it blocks until the video is ready. Without it, the CLI returns a job ID immediately, and you check the status later. This async mode is better when you are submitting multiple jobs in parallel.

Step 5: Try different models

agent-media CLI gives you access to 4 video models and 3 image models. Each model has a unique slug you pass to the generate command. Here are a few examples to try:

To see all available models and their slugs, run:

Bonus: Generate images with Flux 2 Pro

The same CLI also generates images. Flux 2 Pro costs just 5 credits per image and produces photorealistic results in 3-5 seconds. The workflow is identical — swap the model slug and the output is a JPG instead of an MP4.

At 5 credits per image, you can generate 200 images on the Starter plan. Image generation is a great way to prototype visual concepts before committing credits to a full video generation.

Understanding credits

Every generation costs credits. The amount depends on the model and, for video models, the duration. Here is a quick reference for default 5-second video generations and single image generations:

All models are included on every plan. No per-model charges, no separate subscriptions. You can mix and match freely across all 7 models.

What to do next

Now that you have generated your first video, here are some paths to explore:

• *Compare all 7 models — see real benchmarks with generation times, costs, and output samples for every model.
• *Build an automated Shorts pipeline — script batch video generation for YouTube Shorts and TikTok.
• *Turn images into video — learn how to animate still images with Kling and Seedance.
• *Add video generation to Claude Code — set up agent-media as an MCP server for AI-assisted development.