How to Make Videos With Claude Code Using Opus 5: A Step-by-Step Guide

To make videos with Claude Code using Claude Opus 5, you install the Pexo skill, add your API key, and describe the video you want in plain language — the agent then generates a finished result with no editing software, no prompt engineering, and no model selection. Pexo provides a skill you install into Claude Code; Opus 5 is the reasoning model that reads your request, calls the skill, and monitors the job, while Pexo auto-routes each shot across models like Seedance 2.0, Kling 3.0, Veo 3.1, and Sora 2 and returns a finished, multi-shot MP4. Claude Opus 5 shipped on July 24, 2026 as the new default model on Claude Max and the strongest model on Claude Pro, but the important point for this workflow is that the mechanics do not depend on the model version: install the skill, describe the video, review, and export works the same on Opus 5 as it did on Opus 4.8 or Fable 5. This guide walks the full path end to end.

What Claude Opus 5 Changes (and What It Does Not)

Claude Opus 5 is Anthropic's newest Opus-generation model, released July 24, 2026 with a 1,000,000-token context window and API pricing of $5 per million input tokens and $25 per million output tokens. It becomes the default model on Claude Max and the strongest option on Claude Pro, and it adds a toggle that lets you trade cost against capability per task. For video, a stronger orchestrating model means the agent parses a messy, multi-part brief more reliably, keeps a long shot list coherent, and reasons about revisions better. What it does not change is the actual generation: Opus 5 does not itself render pixels. The video comes from the Pexo skill, which routes to dedicated video models. So the workflow is model-agnostic — you get the same finished video whether Claude Code is running Opus 5, Opus 4.8, or Fable 5.

What You Need

Two things, and you can be running in a few minutes:

Claude Code does not generate video on its own, and neither does Opus 5 — the skill is what adds the capability. Claude Code is the most stable host for it because Pexo ships a native SKILL.md there. Once the skill is installed and your key is set, everything else happens inside the conversation. For the different ways an agent can make video, see can Claude Code make videos.

Step 1: Install the Pexo Skill and Add Your API Key

Pexo provides a skill you install into Claude Code — it is not built into the model. Sign in at pexo.ai, copy your API key, install the skill from the open-source repo, then confirm the agent can see it:

# Inside Claude Code (running Opus 5), list installed skills
> /skills
# You should see "pexo" in the list

The Pexo skill ships its helper scripts plus a SKILL.md that tells the agent how to behave — including a rule to pass your request through faithfully rather than rewriting it. Set your API key as the skill instructs (an environment variable), then run the included diagnostic to confirm the config, dependencies, and key are all valid before you start. The skills repo is github.com/pexoai/pexo-skills. Never assume Claude Code or Opus 5 has Pexo "built in" — you install it once, and the same package also runs in Codex, Cursor, and OpenClaw.

Step 2: Describe the Video You Want

This is the whole interface: tell Opus 5 what you want in plain English. You do not pick a model or write a technical prompt. Useful things to specify:

• What it's about — "a product video for these wireless headphones"
• Length and shots — "15 seconds, three shots"
• Mood — "cinematic and premium," "fast-paced for TikTok"
• Music — "ambient electronic," or let the agent choose
• Aspect ratio — "9:16 for Reels," "16:9 for YouTube"

A complete first request looks like this:

> Make a 15-second cinematic product video for these wireless headphones —
  three shots, a slow orbit on the first, premium feel, with ambient music. 9:16.

That is enough. Opus 5 reads the brief, decomposes it into a plan, and hands the job to the Pexo skill.

Step 3: Let the Agent Generate

Once you send the request, Opus 5 dispatches it to Pexo, which runs the full production: it writes a shot script, auto-selects the best model for each shot from 10+ options (a product close-up might route to Kling 3.0, a motion scene to Seedance 2.0, a cinematic wide to Veo 3.1), generates each shot, adds transitions, composes an original score, and mixes a three-layer soundtrack of voiceover, music, and Foley sound effects.

A 15-second, three-shot video takes roughly 8–10 minutes end to end — far faster than choosing models, writing per-model prompts, and assembling clips by hand. While it runs, Opus 5 polls Pexo for progress and reports back; you do not need to do anything until the finished file returns. The larger 1M-token context of Opus 5 helps here because the agent can hold the full brief, the shot list, and the progress log in one conversation without losing track.

Step 4: Review and Iterate

When the video comes back, you review it and ask for changes the same way you described it — in plain language. There is no timeline to edit:

> Make the second shot slower, and swap the music for something more upbeat.

Opus 5 passes the revision through the skill and returns an updated cut. Because the request stays conversational, you iterate by talking, not by re-rendering anything yourself. A stronger reasoning model tends to interpret vague revision notes ("make it feel more energetic") more usefully, but the mechanics are identical on any model.

Step 5: Export and Use It

The finished video comes back as a standard MP4, mastered and ready to post. Because the production is multi-format aware, you can ask for the same content in different aspect ratios — 9:16 for TikTok and Reels, 16:9 for YouTube, 1:1 for feed — without regenerating from scratch. Download it and publish.

Tips for Better Results

A few habits produce noticeably better videos from the same skill, regardless of which Claude model is running:

• Describe the mood, not the model. "Premium and cinematic" guides Pexo's routing better than naming a model — the routing layer translates intent into the right model for each shot.
• Name the platform. Saying "for TikTok" or "for a YouTube pre-roll" sets the aspect ratio, length, and pacing conventions automatically.
• Give 2–4 reference images for products. When accuracy matters — a specific product, logo, or packaging — hand over a few clear photos at 1080p or higher.
• Be explicit about shot count and rhythm. "Three shots, quick cuts" versus "one slow continuous move" produces very different edits.
• Generate variants in the same conversation. Opus 5's long context keeps the brand look consistent across "now a punchier 9-second version" requests in one session.

Other Ways to Start: Image, URL, Script, Audio

Text is only one input. Pexo accepts five, so Opus 5 can start from whatever you already have:

Troubleshooting

Which "Model" Does the Work?

There are two model layers here, and it helps to keep them straight. Claude Opus 5 is the reasoning model inside Claude Code — it reads your brief, calls the Pexo skill, and manages the conversation. The video models — Seedance 2.0, Kling 3.0, Veo 3.1, Sora 2, Runway Gen-4.5, and more — are what Pexo routes to per shot to actually render footage. You never pick either one manually. This is why the tutorial stays valid across Claude releases: upgrading the reasoning model from Opus 4.8 to Opus 5 (or to Fable 5, which Anthropic still recommends for the most advanced autonomous work) improves orchestration, while the video quality comes from Pexo's routing layer, which reshuffles its own model roster every few weeks.

Related reading

• How to Make Videos With Claude Code: A Step-by-Step Guide
• How to Make Videos With Claude Code Using Fable 5
• Can Claude Code Make Videos? The Three Ways, Compared
• Best Video Generation Skills for Claude Code Agents
• How to Make Videos With OpenAI Codex

Resources