Make videos with Claude or ChatGPT
ChatGPT's Sora video tool is discontinued. To make video, add a plug to Claude or ChatGPT.
ChatGPT no longer makes video. OpenAI discontinued the Sora web and app experiences on April 26, 2026 and the Sora API on September 24, 2026. Claude's file creation produces Excel, PowerPoint, Word and PDF files. To make video from either assistant, add a plug: a Remotion skill builds video from code, and generation skills call a paid service. Check what each one costs and where your prompt goes.
Good for
- Marketers who need a short clip from a prompt
- Creators animating a still image
- Developers who want video rendered from React code
- Anyone who first needs a script or storyboard
The short answer
Our advice: use Remotion (code-built video, preview first) for explainers and demos, and a generation skill such as p-video for AI clips. Use the assistants for the script and storyboard. Set a budget before you run paid tools.
Our advice: treat the assistant as the writer and a plug as the camera.
Best for
- Remotion skills for explainers, promos and product demos made from code
- p-video for short AI clips from text or a still image, billed per run
- Claude or ChatGPT for scripts and shot lists
Skip it if
- You need a free, private tool: videoagent sends prompts to a hosted proxy that its SKILL.md does not say who runs
- You expect ChatGPT to still make Sora videos
Go from idea to a finished video clip
- 1
Choose a route
The video skill compares four routes: video built from code, AI generation, AI avatars, and editing or repurposing. Tell it your goal and tools first.
PromptI want a 30-second product demo for [product]. Compare the routes and recommend one for a beginner with a small budget.Uses video
- 2
Write the script and shot list
Let the assistant write the words and the shots before you spend credits on generation.
PromptWrite a 30-second script as a beat table with a hook, three scenes and a closing line. Then list each shot with camera angle and length. - 3
Make the clips
For code-built video, Remotion starts a live preview and only renders a file when you ask. For AI clips, p-video runs Pruna models through the belt CLI and uses inference.sh credits.
PromptGenerate a 5-second 720p clip of scene 1. Show me the price before you run it.Tip: Run a cheap draft first. One p-video model lists $0.05 per 480p clip as of November 2026.
- 4
Add a voiceover
ElevenLabs runs a hosted server you can connect to Claude or ChatGPT. It can generate speech and video, and each job uses your ElevenLabs credits.
PromptCreate a voiceover for this script in a calm voice.Uses ElevenLabs
- 5
Trim and export
The ffmpeg skill gives Claude commands to trim, resize, compress and convert video and audio.
PromptTrim the first two seconds, resize to 1080x1920 and compress under 20 MB.Uses ffmpeg
Best plugs for this
Remotion leads on installs. Generation tools are paid or send your prompt off your machine.
- 1remotion-best-practicesSkill
Remotion's own skill. Builds video from React code, shows a live preview and renders only when you ask.
727,896 installs - 2videoSkill
Helps you choose between code-built video, AI generation, avatars and screen recordings before you spend money.
90,005 installs - 3p-videoSkill
Generates short videos from text or an image with Pruna models. One model lists $0.05 per 480p clip; each run uses inference.sh credits.
13,108 installs - 4videoagent-video-studioSkill
Makes short AI videos from text or images. Needs no API key by default, but sends your prompt to a hosted proxy that its SKILL.md does not say who runs.
10,484 installs - 5
ElevenLabsOfficialHostedOAuthElevenLabs' own hosted server for Claude or ChatGPT: speech, images and video. Spends your ElevenLabs credits.
- 6ffmpegSkill
Commands to convert, trim, resize and compress video and audio.
12,423 installs
ChatGPT
ChatGPT no longer makes video. Use it for scripts, images and plugs.
Plan: Any plan for images and scripts. Video needs a plug.
What it can do
- Create and edit images with ChatGPT Images, available on all tiers
- Connect ElevenLabs' hosted server for speech and video
- Export your old Sora content at sora.chatgpt.com/sunset
- Share live video or your screen in Advanced Voice Mode (eligible subscribers). This is input to the chat, not video creation
Limits
- The Sora web and app experiences were discontinued on April 26, 2026
- The Sora API was discontinued on September 24, 2026
Claude
Claude is the scriptwriter and the driver of video tools.
Plan: Any plan for scripts and storyboards. Video skills that run commands need Claude Code or another coding agent.
What it can do
- Create Excel, PowerPoint, Word and PDF files, such as a script or deck
- Build diagrams, charts and interactive visuals in chat (beta on web and desktop)
- Run video skills such as Remotion and p-video in Claude Code
- Connect ElevenLabs' hosted server for speech and video
Limits
- File creation covers Excel, PowerPoint, Word and PDF files
- Generation plugs use paid credits from their own providers
Questions
Can ChatGPT make videos?
Not any more. OpenAI discontinued the Sora web and app experiences on April 26, 2026 and the Sora API on September 24, 2026. ChatGPT Images still makes and edits pictures.
Can I still get my Sora videos?
OpenAI says you can export them at sora.chatgpt.com/sunset by clicking Export. Data is permanently deleted after the discontinuation and any final export window, so do it soon.
Does ChatGPT have any video feature now?
Only live input. OpenAI's release notes (July 8, 2026) say eligible subscribers can share video or their screen in Advanced Voice Mode. That is camera and screen input to the chat, not video creation.
Can Claude make videos?
Claude's file creation covers Excel, PowerPoint, Word and PDF files. With a plug such as a Remotion or p-video skill in Claude Code, it can build or generate video.
What is the cheapest way to make AI videos?
Our advice: start with the Remotion skill, which builds video from code and starts with a local preview. Check Remotion's licence terms for your use. videoagent says it needs no API key by default, but it sends prompts to a hosted proxy that its SKILL.md does not say who runs. Generation tools such as p-video use credits.
How much does an AI video cost?
It depends on the tool. As of November 2026 the p-video skill lists $0.05 to $0.11 per short clip on its WAN models and per-second pricing for P-Video. ElevenLabs jobs use your ElevenLabs credits.
Can I turn a photo into a video?
Yes, with a plug. The image-to-video and p-video skills animate a still image through inference.sh and use credits on each run.
Can I make a talking avatar video?
Yes. The p-video-avatar skill turns a portrait image into a talking video with built-in voices or your own audio, and uses inference.sh credits.
Can I use these plugs in ChatGPT?
The ElevenLabs server connects to ChatGPT as well as Claude, and ElevenLabs lists video generation among its tools. The video skills run commands such as belt and npx remotion in your shell, so use them in Claude Code or another coding agent.
Is it safe to put my script or photos into these tools?
Generation tools send your prompt and files to a third party. inference.sh uploads local files you reference, and videoagent posts prompts to a proxy. Avoid private text and images.
Sources
Docs, issue trackers and discussions this page is built from.
Show all 10Hide
- What to know about the Sora discontinuation · help.openai.com
- Create and edit files with Claude · support.claude.com
- p-video SKILL.md · github.com
- videoagent-video-studio SKILL.md · github.com
- ElevenLabs hosted MCP server · elevenlabs.io
- ElevenLabs MCP · elevenlabs.io
- ChatGPT release notes · help.openai.com
- Images in ChatGPT · help.openai.com
- Visual and interactive content · support.claude.com
- Using Skills in Claude · support.claude.com