OpusClip

• Turning a two-minute product update filmed on a phone into a captioned, B-rolled vertical video without opening an editor
• Matching the caption, motion graphics, and transition style of a reference TikTok or YouTube video on your own talking-head footage
• Saving a weekly FAQ or tips video setup as a Recipe so every new recording gets the same captions, hook, and branding
• Revising a cut from inside Claude or ChatGPT by asking for specific changes such as removing the music or trimming the intro
• Clipping a long podcast or webinar into shorts, then scheduling them to YouTube, TikTok, Instagram, and LinkedIn from one place
• Having Codex or Claude Code edit a raw take through the AI Producer plugin, where the account has access

OpusClip's move from clipping into editing targets the people who film a quick take on their phone and never get around to cutting it. AI Producer is narrow on purpose: one person talking to camera, up to five minutes, one finished MP4. Inside that scope, the brief-plus-revisions loop and reference-style matching cover most of what a short-form editor would do, and being able to run it from Claude or ChatGPT suits people who already work there. Recipes and the existing clipping and scheduling tools make it more useful for a recurring format than for one-off edits. The weak spots are cost visibility, since AI Producer bills credits in slices without a published rate, and scope: interviews, multi-camera shoots, screen recordings, and long-form edits still need Descript, invideo, or a traditional editor.

About OpusClip

OpusClip started as a clipping tool that finds the best moments in a long video and turns them into captioned vertical shorts. OpusClip 4.0, announced on November 2, 2026, moves it into editing raw footage, with three pieces that work together. Auto-edit takes a raw talking video or a link and applies the edits you switch on (speech cleanup, captions, a Viral Hook, motion graphics, B-roll) to return a finished video. Recipes save that setup, including aspect ratio, which edits run, their styles, and a Brand Kit, so the next video is edited the same way; OpusClip also ships built-in Recipes such as Auto-edit talking videos, Clip long video, Clean up recording, and Translate & dub. AI Producer is the agentic part. You upload a video of one person talking to camera, optionally add a reference (a style preset, an Instagram, TikTok, or YouTube link, or an uploaded video or images), and describe the edit in your own words. The agent reads the footage and transcript, plans the cut, and applies its editing skills: rough cut, voice cleanup, captions, motion graphics, camera movement, auto layout, auto B-roll, floating overlays, and, in beta, sound effects, background music, and scene restyle. Each follow-up request (bigger captions, no music, cut the intro) updates the same project rather than starting over, and the export is one MP4 at up to 1080p and 30 fps in 9:16, 16:9, or 1:1. The editing conversation can also run in Claude through the AI Producer connector or in ChatGPT through the AI Producer plugin, where the model that plans the edit runs on your Claude or ChatGPT plan; Codex, Claude Code, and Cursor Desktop can install an MIT-licensed plugin with editing skills and an MCP connection. Separately, a hosted OpusClip MCP server and REST API let agents and developers clip, caption, reframe, and schedule video. Everything draws on one OpusClip credit balance. The main tradeoffs are scope and cost visibility. AI Producer is built only for single-speaker footage of up to 5 minutes, with no multi-camera or screen-recording support yet; the coding-agent path is open to a limited set of accounts; and AI Producer charges credits in slices as it works without a published per-minute rate. Compared with Descript or invideo, OpusClip is less a general-purpose editor than a fast path from a phone-shot take to a posted short.