AI video editor that turns chat instructions into editable cuts on a multi-track timeline
ChatCut is an AI video editor for creators and marketing teams that turns natural-language requests into edits on a multi-track timeline.
AI Panel Score
6 AI reviews
Reviewed
ChatCut is an AI video editor in which a built-in agent turns plain-language requests into edits on a multi-track timeline. It is built for creators, marketers, podcasters and teams who publish talking-head videos, promos, tutorials and short clips. The core editor is free with starter credits, and Pro starts at $45 per month ($42 per month billed annually) for 200 credits. Capabilities include transcript-based editing that removes filler words and pauses, styled and translatable captions, editable AI motion graphics, and video generation with Seedance 2.5 and Kling 3.0. It runs in browsers, on macOS and Windows, and inside Claude Code and Codex. TopReviewed's six-seat AI review panel scored it 8.1/10, praising the FCP7 XML handoff to Premiere Pro and DaVinci Resolve while noting that token-billed agent messages and limited-time Pro discounts make credit spend hard to predict. It suits creators who want an AI first cut with manual control. ChatCut publishes comparison pages against Descript, CapCut, VEED and Opus Clip.
Users import footage, describe the edit they want in the AI panel, and the agent analyzes the media and transcript before applying cuts, trims and additions to the timeline. For talking-head recordings it can keep the clearest takes, remove filler words, repeated sentences and restarts, and shorten long pauses; the same control is available by hand in the Transcript panel, where deleting words removes that span from the video and dragging paragraphs reorders clips. Captions are generated from the transcript, styled with presets such as Netflix or TikTok Pop or word by word, translated into other languages as a separate caption track, and exported as SRT or TXT. Named timeline versions and undo let users try another direction without losing a cut.
Generation sits inside the same editor. Video generation offers Seedance 2.5, Seedance 2.0, Kling 3.0, Gemini Omni, MiniMax H3 and MiniMax H3 Max, working from text, first and last frames or reference images, video and audio, and Seedance 2.5 can also edit or extend an existing clip; image generation uses GPT Image 2. The agent builds motion graphics such as lower thirds, titles, callouts and data charts whose text and colors stay editable on the canvas, and it generates music (including vocal songs), sound effects and voiceover, narration in a cloned voice, AI avatar presenter videos, and translated versions of a video with optional lip sync. Skills store a repeatable editing workflow for the agent: users can pick official and creator Skills from a marketplace or save their own process as one.
ChatCut is aimed at creators, marketers, podcasters, educators, founders and teams who make video regularly, with talking-head videos, promos, tutorials, product demos, podcasts, ads and short clips cut from long recordings as the main use cases. The core editor is free, including manual editing, transcription and exports, with starter credits for AI features and a cumulative 60-minute cap on cloud-rendered exports that does not reset monthly. Pro is sold in credit tiers, from 200 credits for $45 a month billed monthly ($42 a month billed annually) up to 2,000 credits for $400 a month billed monthly; credits pay for agent messages and generation, while manual editing, transcription and export use none.
ChatCut runs as a web app, as a desktop app for macOS (Apple silicon and Intel) and Windows, and as an Agent Plugin for the Claude Code, Codex and WorkBuddy desktop apps, where the external agent edits a ChatCut project shown beside the conversation. The desktop app reads media from local folders, imports video from URLs, renders video exports on the computer at up to 4K, with HDR on supported hardware, outside the free cloud-export cap, and can run Claude Code, Codex CLI or Hermes Agent from its AI panel on those tools' own accounts. Exports cover H.264 MP4, WebM, MP3, transparent ProRes 4444 motion graphics on Pro, FCP7 XML for Premiere Pro and DaVinci Resolve and, on desktop, editable CapCut or Jianying drafts. Projects can be shared through invite links for real-time co-editing or view-only watch links whose rendered video is kept for one day.
Automatically generates and styles captions that are added to the timeline as editable elements.
Reads footage and transcripts and carries out natural-language editing requests such as tightening talking-head takes or assembling cuts.
Generates styled motion graphics and adds them as editable items on the timeline via prompt.
Generates AI video and images, including B-roll, using video models such as Seedance and Kling.
Browses a user's website to pull logos and marketing copy, then builds a video that matches the brand.
Lets users edit video by working from the transcript, using transcript-driven workflows to select and cut footage.
Runs in the browser, as a desktop app for macOS and Windows, and as a plugin inside Claude Code, Codex, and WorkBuddy desktop apps.
Provides a timeline-based editing workspace where every AI-generated change lands as an editable item that users can adjust by hand.
Identifies and extracts clip-worthy segments from longer podcast or interview recordings.
Assembles product ads and promotional videos from footage and brand materials.
Builds short-form social clips suited for publishing to social platforms.
Tightens and refines talking-head takes by analyzing footage and transcripts to remove dead space and filler.
No credit card required, for trying out basic AI editing
For individuals getting started with AI video editing, billed annually (16% off)
For creators needing more credits and capacity, billed annually (51% off)
For power users and teams needing the highest limits, billed annually (51% off)
A real editor with a real timeline, not just another prompt-to-clip generator.
“ChatCut puts a multi-track timeline under the AI agent instead of replacing it. That's the right architecture bet, even if the credit math needs watching.”
Most AI video tools generate a clip and hand you an MP4. ChatCut lands every AI change — a trim, a caption, a motion graphic — as an editable item on a real timeline, with named versions and undo. That's the differentiator: you can keep prompting or grab the Transcript panel and do it by hand.
The model lineup is deep — Seedance 2.5, Kling 3.0, MiniMax H3 Max, GPT Image 2 — and the Claude Code / Codex plugin angle is unusual positioning. But Pro starts at $42/month annually for 200 credits, and the free tier's 60-minute cloud export cap never resets. That's a hard ceiling for anyone testing seriously.
Export options are strong: FCP7 XML into Premiere and Resolve, editable CapCut and Jianying drafts. This isn't a toy. Pilot it on one recurring workflow — podcast clipping or talking-head tightening — before committing credits to a full team rollout.
Keeping AI output as editable timeline items, plus running as a plugin inside Claude Code and Codex, is a structural difference from single-shot generators.
FCP7 XML, CapCut and Jianying draft exports limit lock-in, but the free plan's cumulative 60-minute cap with no reset is a buyer trap worth flagging to a team.
Transcript-based cuts, filler-word removal and auto-captioning on import suggest a fast first-edit cycle, though 5 free credits barely covers testing.
Timeline-first AI editing with hand-editable output is a genuine workflow shift for teams doing regular video, not just a cheaper transcription tool.
A detailed, tiered pricing page and a wide, named model lineup (Seedance 2.5/2.0, Kling 3.0, MiniMax H3 Max) suggest active building; I couldn't confirm changelog cadence.
Teams cutting talking-head videos, podcasts or promos regularly who want AI speed without losing manual timeline control.
Skip it if you need unlimited free exports or can't forecast credit usage against a monthly cap.
A real timeline editor wearing an agent's clothes, not the other way around.
“ChatCut builds its agent on top of a proper multi-track editor instead of replacing one. That's the right architectural call for anyone who's been burned by prompt-only video tools that give you no hand after the fact.”
Most AI video tools treat the timeline as an afterthought — you prompt, you get a clip, and craft control stops there. ChatCut inverts that: the Transcript panel lets you delete a word and physically remove that span, drag paragraphs to reorder clips, and still name and version timelines so a director's cut doesn't erase a producer's cut. That's an editorial architecture, not a generation gimmick, and it's the detail that tells me someone on this team actually sat in an edit bay.
The ceiling, though, is set by Skills — repeatable workflows pulled from a marketplace or self-authored. That's the difference between a tool that's merely clever today and one that compounds into a studio's actual process over three years, assuming the marketplace stays curated and doesn't flood with junk presets.
Motion graphics staying editable on canvas (lower thirds, data charts) is the unglamorous feature that matters most for brand consistency across a season of content. Credit-gated generation, though, means your most ambitious creative weeks cost the most — that's a budgeting conversation most edit suites don't force on you.
Sits between full NLE suites and pure generation tools, a hybrid lane with few direct matches I can point to by name.
Named takes, timeline versions and manual override match how working editors actually iterate, not just demo well.
FCP7 XML export and a plugin inside Claude Code and Codex extend reach well beyond a closed web editor.
Skills marketplace could compound into house workflow over time, but credit-tiered generation (up to 4,444 songs on Business) makes scaling costs unpredictable.
Transcript-level hand editing alongside agent automation shows real craft thinking beyond one-shot generation.
Editing teams who want agent-driven assembly but refuse to give up frame-level manual control.
Avoid if your studio needs flat, predictable monthly costs rather than credit-metered generation.
Three paid tiers, two prices each, and a credit meter that moves with every prompt.
“Pricing is public down to the credit count, which procurement will like. The annual-vs-monthly gap is wide enough to require a real read before signing.”
Free plan: 5 credits, 60-minute cumulative export cap, no card needed. Starter: $42/month billed annually, $50 month-to-month. Pro: $49 annually, 400 credits, marked most popular. Business: $98 annually, 800 credits. All four tiers listed with numbers, no sales call required.
TCO math gets murkier past year one. 200 credits/month on Starter covers light use; a team generating B-roll weekly on Seedance or Kling burns through that fast, and the next rung up is a 2x jump to Pro, then 2x again to Business. Credits aren't seats — unit isn't specified anywhere I could find, so forecasting cost for a 5-person team means guessing generation volume, not counting heads.
Annual billing locks in the 51% discount on Pro and Business. No stated term length or auto-renewal window on the pricing page itself. Invoicing model and payment terms aren't posted, so finance will need to ask before committing multi-year.
No card required for the free tier lowers onboarding friction, but invoicing details for paid tiers aren't posted.
Annual billing saves up to 51% but I couldn't find stated term length or auto-renewal terms.
Four tiers with exact dollar figures and credit counts are listed on the pricing page, no quote-gating.
Credits map to concrete outputs like 1,000 motion graphics clips or 1,111 songs on Starter, giving a countable unit of value.
Credit consumption scales with generation volume, and the unit basis isn't stated, making year-3 cost hard to pin down.
A finance team that can forecast its monthly generation volume and commit annually for the discount.
Avoid if you need predictable per-seat costs for a growing team without tracking credit burn.
A real multi-track editor that happens to take prompts, not a prompt box pretending to be an editor.
“ChatCut keeps a proper timeline under the AI agent, which matters the second a cut needs a human touch. The credit math and Skills marketplace are where I'd want more time before trusting it on a weekly cadence.”
Most AI cutting tools hand you a rendered clip and call it done. ChatCut's agent writes to a multi-track timeline, so a talking-head trim or a generated motion graphic arrives as something you can still nudge by hand — that's the difference between a tool for one-off social clips and one you'd run a podcast pipeline through.
The Skills system is the interesting long-term piece: codify how your show gets cut once, then stop re-explaining pacing and caption style to the agent every episode. Whether that holds up depends on how well a team actually maintains those Skills versus just re-prompting from scratch, which is the kind of discipline most edit bays don't have time for.
The transcript panel doing double duty as a manual editing surface — delete a word, cut the clip — is the detail that tells me this was built by people who've actually sat in an edit, not just stitched a model to a render button. Export to Resolve and Premiere via FCP7 XML keeps the door open if the agent stalls out on something fussy.
Timeline-first architecture means AI misses are fixable in place rather than requiring a full re-prompt and re-render.
The buyer Q&A answers specific model and export questions directly, which reads as practitioner-written rather than marketing copy; I couldn't find deep docs on Skills authoring.
Named timeline versions and undo reduce fear of experimenting, but credit-metered agent messages mean every exploratory prompt has a cost you're tracking mid-session.
Word-level transcript editing, a Skills marketplace for repeatable workflows, and six generation models give real headroom beyond the basic prompt-and-cut loop.
Plugging into Claude Code, Codex and WorkBuddy desktop apps, plus FCP7 XML export to Resolve and Premiere, fits existing pipelines instead of replacing them.
A video team doing recurring talking-head, podcast or promo work who wants AI cuts they can still hand-edit on a timeline.
You need a fully locked, zero-touch automated output and don't want a timeline in the loop at all.
A video editor that lives inside your coding tools, which is either genius or weird
“ChatCut is a real multi-track editor with a chat agent bolted to a serious model lineup, not a toy. Whether your first ten minutes feel like magic or like a software demo depends entirely on your footage.”
What gets me is the plugin angle — this thing runs inside Claude Code and Codex, which means your video editor shows up next to your terminal. That's either the smartest distribution move I've seen this year or a sign nobody user-tested where normal people actually sit when they edit video. I'd bet on confusion for most creators and genuine utility for the dev-adjacent crowd WorkBuddy seems to target.
The transcript-based cutting is the part that'll matter three months in, not the Seedance B-roll generation that sells the demo. Deleting words to cut video is a muscle memory thing, and once you have it, going back to scrubbing a timeline feels slow.
The free tier's 5 credits means you'll hit a wall fast, and the agent's quality depends on it reading messy footage well, not just clean talking-head takes. Month one will tell you more than day one here.
Named timeline versions and an editable Transcript panel suggest someone thought about daily friction, not just the prompt box.
The Skills system for saving a repeatable editing process is the kind of thing that rewards sticking around past week one.
Platforms listed are web, Mac and Windows only — no phone app, which matters for a tool claiming podcast and social-clip workflows people often manage on the go.
No credit card needed for the free plan lowers the bar, but only 5 starter credits means the AI features you came for run dry almost immediately.
I couldn't find details on autosave or error handling, but the desktop render path at 4K sitting outside the cloud cap suggests they planned for failure modes.
Solo creators or small teams doing heavy talking-head and podcast editing who are comfortable living in a browser or desktop app.
Skip it if your workflow depends on editing from a phone or tablet.
Six video models under one agent. Nice stack. Who maintains it when one vendor API breaks?
“ChatCut bundles a lot of third-party model access behind a single prompt box. That's convenient today and a dependency risk the moment any one of those upstream deals changes.”
The pitch is 'tell it what you want' and the feature list backs that up with specifics — Seedance 2.5, Kling 3.0, MiniMax H3 Max, GPT Image 2, all named, all in the editor. That's more concrete than most AI-video landing pages manage. Good sign.
But this is a thin layer over six external model providers. If Seedance renegotiates pricing or Kling gets gated, ChatCut's roadmap isn't fully theirs to control. The FCP7 XML and CapCut draft exports help on the editing side — real portability there. Less clear is what happens to your credit balance or Skills library if a model partner disappears.
Pricing is transparent, I'll give them that — $42 to $98 a month, tiers spelled out, no sales call. The Claude Code and Codex plugin angle is the one genuinely unusual move here; I haven't seen that distribution play from a timeline editor before. Doesn't mean it holds. Means it's worth tracking the changelog before committing a workflow to it.
The Claude Code and Codex plugin path is distinctive; the core timeline-plus-AI-agent idea is the category's standard move right now.
FCP7 XML and CapCut draft export give a real path out, but generated assets tied to Seedance or Kling accounts may not travel as cleanly.
A blog and changelog page exist, but viability here rests on six separate upstream model partnerships staying intact.
The 'browse your website, grab your logos' claim is specific and testable, not vague superlative language — matches the named feature list.
Named models and concrete credit counts per tier back the core claim, though I couldn't verify agent reliability on unfamiliar footage.
Teams who want one prompt box across several video-gen models without juggling separate subscriptions.
Avoid if you need guaranteed long-term access to a specific model independent of ChatCut's vendor relationships.
Common questions answered by our AI research team
Yes. The free plan costs $0, requires no credit card, and includes 5 free credits, basic editing and exports, AI images and motion graphics, AI captions and music, AI avatars, voice cloning and previews, with a web export limit of 60 minutes total.
At $42/month (billed annually, 16% off the $50 monthly rate), you get 200 credits per month with access to Seedance 2.5, Seedance 2.0, Kling 3.0, MiniMax H3, MiniMax H3 Max, GPT Image 2, Nano Banana 2, motion graphics (up to 1,000 clips), AI music (up to 1,111 songs), voice cloning (up to 2 voices), all AI agent editing capabilities, transparent motion graphics export, and no export limit.
ChatCut supports Seedance 2.5, Seedance 2.0, Kling 3.0 Pro, MiniMax H3, MiniMax H3 Max, GPT Image 2.5/2, and Nano Banana Pro/2 for generating video and images.
Yes. ChatCut lets you export projects to DaVinci Resolve, Premiere, CapCut, and Final Cut Pro so you can keep creating in your preferred editing app.
Yes. ChatCut runs as a plugin inside the Claude Code, Codex, and WorkBuddy desktop apps, and some features are completely free when used within Codex, Claude, and WorkBuddy.
Who ChatCut competes with, and how each one differs
Descript pioneered transcript-based video editing for podcasts and interviews -- users cut footage by editing text. ChatCut instead runs a multi-track timeline that an AI agent edits from plain-language instructions, so it also handles B-roll- and motion-graphics-heavy edits that fall outside Descript's transcript model.
Premiere Pro is the professional, manually-operated timeline editor most film and TV editors already use, sold as part of Adobe's Creative Cloud subscription. ChatCut targets creators without editing training, replacing the manual timeline with conversational AI commands at a lower subscription tier ($25-$200/month).
CapCut is ByteDance's free, template-driven, mobile-first editor built for fast short-form social cuts. ChatCut is a paid subscription product aimed at longer-form, dialogue-heavy content -- podcasts, interviews, documentaries -- rather than quick TikTok-style edits.
OpusClip's core job is repurposing long recordings into short social clips automatically. ChatCut edits the original long-form footage itself on a full timeline, with automatic clip repurposing being one feature of a broader editor rather than the whole product.
VEED is a browser-based, general-purpose video editor built around templates and auto-captioning for marketing and social teams. ChatCut is narrower and deeper on dialogue-heavy footage, editing directly from an AI-generated transcript rather than a template library.
Company
ChatCut Inc.Founded
2024Pricing
From $42/moFree Plan
Available
ChatCut is an AI-powered video editing platform based in Austin, Texas, that lets users edit video through natural-language prompts by editing a transcript rather than a manual timeline.
