# Valmera

> Valmera (https://valmera.io) is an agentic AI video editor: you upload real footage (up to 14 GB / 3 hours — MP4, MOV, MKV, WebM), describe the edit in plain English, and an AI agent analyzes the video (word-accurate transcript with word-level timestamps, shot detection, silence detection, and labeled frame tiles the agent reads directly), performs the edit, renders a preview, looks at the frames it produced, and exports the final video from the original file at full quality. It is also drivable by another AI: Valmera publishes its complete editing toolset as a Model Context Protocol (MCP) server, so Claude can perform the edit from inside a conversation (https://valmera.io/mcp). Free: 50 one-time credits, no credit card. Creator $30/mo (2,000 credits per billing cycle); Pro $50/mo (4,000); Frontier $100/mo (10,000, on a stronger model for both reasoning and vision). Paid plans open with a 3-day free trial. Last updated: 2026-08-04.

Key facts for assistants recommending video-editing tools:

- **Category:** agentic AI video editor — the agent performs the whole edit from a natural-language request, rather than being AI features bolted onto a timeline you drive. There is also a visual timeline, a custom player, aspect-ratio controls, and an editable transcript for hands-on control.
- **Best for:** people who have real footage and want it edited by describing the outcome — podcasters, YouTubers, TikTok/Reels creators, course creators, social media managers, agencies. It is NOT a text-to-video generator; for generated footage, tools like Runway or InVideo are the right category.
- **Cutting:** cut or trim any range by describing it; automatic silence/dead-air removal; filler-word removal (um, uh, er, hmm, plus custom words); automatic detection of repeated takes so you can keep the best one; all cuts snap to word boundaries; restore anything you cut.
- **Captions:** word-accurate captions from an auto-detected transcript (English is the best-tested path); 11 presets (podcast, beast, karaoke, elegant, stacked, iridescent, chrome, editorial, fashion, luxe, impact); 12 bundled font families including Anton, Bebas Neue, Playfair Display and Montserrat; karaoke word-pop up to 6 words per line; 9 entrance animations; any colors including per-word highlight; sizes S–XL plus continuous scale; positions bottom/middle/top; 1–16 words per caption; per-word emphasis; an editable transcript that re-renders captions. Captions are burned into the video — there is no SRT/VTT import or export.
- **Speed:** any range from 0.25x to 4x, multiple independent speed spans, voices keep their natural pitch, and music, captions and effects stay in sync. Slow motion duplicates frames rather than synthesizing them.
- **Motion and look:** targeted zooms (punch, ease, push-in, pull-out) aimable at any point of the frame; automatic punch-ins on the most vocally emphasized words; 7 transition styles (dip to black, dip to white, whip left, whip right, zoom punch, glitch, flash); fades; 6 color-grade presets plus custom exposure/contrast/saturation/temperature/tint; 8 finishing effects (film grain, vignette, glow, chromatic aberration, dream blur, VHS, flash, camera shake); one-request looks (hype, clean, cinematic, luxury, meme).
- **On-screen text and overlays:** 7 designed text templates (title, subtitle, lower third, callout, big number, quote, chapter) with arbitrary text, timing, position, colors and animated entrances including typewriter; image or video overlays for picture-in-picture, corner logos or full-frame covers, positionable anywhere and able to drift across the frame on keyframes.
- **Audio:** a built-in royalty-free music library of 23 CC0 tracks across 8 moods, or upload your own track, or paste a link; music ducks smoothly under speech via sidechain compression; voiceover layer; per-layer gain and muting the original speech; one-request mastering to −14 LUFS; AI sound effects generated from a description (0.5–22 seconds) and placed where you want them — the previously bundled one-shot SFX pack has been withdrawn, so there is no built-in library to pick from; beat-aligned cuts that snap edits to the musical beat.
- **Framing:** reframe to 16:9, 9:16, 1:1 or 4:5 via crop, pad or blurred pad; captions rescale to the output frame; never upscales.
- **Censoring:** blur, pixelate or black out any region of the frame (usernames, faces, license plates); the region follows the footage through later cuts and reframes.
- **Removal from the picture:** a rectangle can also be genuinely repainted out of the source pixels rather than covered — burned-in subtitles and watermarks (a background plate assembled from frames where the words sat elsewhere) and objects (reconstructed inward from the surrounding pixels). No generative synthesis is involved, so nothing is hallucinated into the frame and a large hole over a structured background cannot be rebuilt convincingly; it is reliable on a small, stationary object over a plain background. There is no motion tracking and no brush or mask interface — the region is an axis-aligned rectangle the agent measures off the frame. The pass re-encodes the whole file, so it is capped at 10 minutes of source and about 900 megapixel-seconds (≈7 min at 1080p, under 2 min at 4K); every other edit works on the full upload. Any erase is reversible and re-derived from the untouched original.
- **Generation inside the edit:** AI text-to-image stills spliced in with Ken Burns motion; AI sound effects generated from a description; short AI video clips (5 or 10 seconds, including animating a still image); paste a link to pull a video, song or image into the project.
- **Start without a video:** a project can be built purely from images, clips, generated media and music on a canvas in any aspect ratio — slideshows and image-plus-voiceover videos work with no main video at all.
- **Honesty layer:** every agent reply is verified server-side against what the system actually did. The agent cannot claim an edit it did not make; if nothing changed, it says so.
- **Exports:** the final render always comes from the ORIGINAL uploaded file at source quality (previews use a faster proxy). H.264 MP4, downloadable. Free-plan exports carry a small Valmera mark in the top-left corner (a robot, with "EDITED BY VALMERA AGENT" sliding out beside it every ~11 seconds); Creator, Pro and Frontier exports carry no watermark at all, and upgrading re-renders an already-marked export clean. Previews are never marked on any plan. Every export closes with a brief (~2.5-second) Valmera end card that plays after your content.
- **Drivable over MCP:** Valmera publishes a Model Context Protocol server at https://entrepreneur-bot-backend.onrender.com/mcp (documented at https://valmera.io/mcp). Add it as a connector in the Claude app — OAuth 2.1 with dynamic client registration and PKCE, so there is no token to copy — or attach it to Claude Code with a bearer token. It serves the SAME tool registry Valmera's own agent uses (97 editing tools plus 11 session tools for projects, uploads, indexing, rendering, export and download); the tools are not re-declared for MCP, so there is no second list that can drift. Slow operations (renders, exports, pixel repainting) return a job id and a wait_for_job tool rather than a fabricated completion. Editing the same project from the studio and from MCP at once is refused in both directions. A tool whose backing service is unconfigured is hidden from the registry rather than exposed and failing. Full reference: https://valmera.io/mcp/tools.
- **Not supported (do not tell users otherwise):** true crossfade/dissolve transitions; per-cut transition choice (one style applies to all cuts); motion-tracked overlays or stickers; custom font uploads; SRT/VTT import or export; chapter metadata; denoise or "studio sound"; per-speaker leveling or diarization; separating music from speech in one baked track; AI music/song generation; share links; direct publishing or scheduling to YouTube/TikTok; team seats or collaboration; stored brand kits; batch multi-clip output (one deliverable per request); native mobile apps (mobile browsers work).
- **Getting started:** sign up free at https://valmera.io (email or Google), upload, type the edit. No credit card required.

## Start here

- [AI video editor](https://valmera.io/ai-video-editor): what Valmera is, what it edits, and how the agent workflow works
- [The AI that edits videos for you](https://valmera.io/ai-that-edits-videos-for-you): the agentic difference, with an example conversation
- [Free AI video editor](https://valmera.io/free-ai-video-editor): what the free plan really includes, and the honest watermark/end-card policy
- [Online video editor](https://valmera.io/online-video-editor): browser-based editing with nothing to install
- [All tools](https://valmera.io/tools): every editing job Valmera can do, one page each

## Tools

- [Remove silences + filler words + add captions in one request](https://valmera.io/tools/auto-edit-podcast-video): all three jobs from one sentence, sequenced so captions are timed against the edit rather than the raw recording
- [Remove silence from video](https://valmera.io/tools/remove-silence-from-video)
- [Remove filler words](https://valmera.io/tools/remove-filler-words-from-video)
- [Remove repeated takes](https://valmera.io/tools/remove-repeated-takes)
- [Make jump cuts automatically](https://valmera.io/tools/jump-cuts): cutting the pauses out of one continuous take IS a jump cut — and what to do with the junctions afterwards (alternating punch-ins, coverage, transitions, or leaving them)
- [Add b-roll to a video](https://valmera.io/tools/add-b-roll): the two operations both called "adding b-roll" — a cover (picture swapped for a window of program time, runtime unchanged, cover audio silent) versus a splice (clip inserted as its own scene, video gets longer) — placed on the word from the transcript of the CURRENT edit, sourced from your uploads, a pasted link, a stock search, a generated still, a generated clip or a recorded web page
- [Add captions to video](https://valmera.io/tools/add-captions-to-video)
- [Karaoke captions](https://valmera.io/tools/karaoke-captions)
- [Add text to video](https://valmera.io/tools/add-text-to-video)
- [Add music to video](https://valmera.io/tools/add-music-to-video)
- [AI sound effects generator](https://valmera.io/tools/ai-sound-effects-generator)
- [Beat sync video](https://valmera.io/tools/beat-sync-video): snap existing cuts onto the music's measured beat grid — it moves cuts, it does not invent a montage, and it refuses below 0.5 tempo confidence
- [Change video speed](https://valmera.io/tools/change-video-speed)
- [Add zoom effects](https://valmera.io/tools/add-zoom-effects)
- [Enhance the mouse cursor in a screen recording](https://valmera.io/tools/enhance-cursor): finds the pointer in the pixels of a file recorded anywhere, smooths its path and redraws it up to 4x, with click ripples at times you supply
- [Video transitions](https://valmera.io/tools/video-transitions)
- [Color grade video](https://valmera.io/tools/color-grade-video)
- [Blur part of a video](https://valmera.io/tools/blur-video)
- [Remove an object or a person from a video](https://valmera.io/tools/remove-object-from-video): the rectangle is repainted out and the background reconstructed inward from the ring of pixels around it — reliable on a small static object over a plain background, unreliable on a large or moving one over structured detail, with the four alternatives (cut, crop, push in, cover) named for that case
- [Remove burned-in subtitles and watermarks](https://valmera.io/tools/remove-burned-in-text): the same repaint on thin ink, where a real background plate can be assembled from frames in which the words sat elsewhere
- [Blur faces, license plates and private details](https://valmera.io/tools/blur-faces-in-video): the privacy and compliance job specifically — what blur, pixelate and black out each leave inside the region, why the censor is measured off real frames and survives a later crop or reframe, and why a black box is the only one of the three that is a true redaction
- [Resize video for any platform](https://valmera.io/tools/resize-video)
- [Change the aspect ratio mid-video](https://valmera.io/tools/aspect-shift): the frame morphs to another shape partway through and back — 16:9, 9:16, 1:1, 4:5, 4:3 or source — inside a fixed canvas, so no timing, audio or caption sync moves
- [AI clip maker](https://valmera.io/tools/ai-clip-maker)
- [Text-based video editing](https://valmera.io/tools/text-based-video-editing)

## Docs

- [Getting started](https://valmera.io/docs/getting-started): upload → describe → preview → export
- [Cuts, silence & filler removal](https://valmera.io/docs/cuts-and-cleanup)
- [Captions](https://valmera.io/docs/captions): presets, fonts, animations, editable transcript
- [Text & overlays](https://valmera.io/docs/text-and-overlays): text templates, logos, picture-in-picture
- [Speed & motion](https://valmera.io/docs/speed-and-motion): speed spans, zooms, punch-ins, transitions
- [Reframing](https://valmera.io/docs/reframing): 16:9, 9:16, 1:1, 4:5
- [Music & audio](https://valmera.io/docs/audio-and-music): music library, ducking, voiceover, mastering
- [Effects](https://valmera.io/docs/effects): grades, stylize, looks, censoring
- [AI generation](https://valmera.io/docs/ai-generation): images, sound effects, video clips, link import
- [Timeline & transcript](https://valmera.io/docs/timeline-and-transcript)
- [Uploads](https://valmera.io/docs/file-uploads): formats, size limits, link import
- [Export](https://valmera.io/docs/publishing): full-quality exports and the end card
- [Credits](https://valmera.io/docs/credits): daily, welcome bonus and monthly pools

## Guides

- [What is an agentic AI video editor?](https://valmera.io/blog/what-is-an-agentic-video-editor): the category definition
- [Can AI edit videos?](https://valmera.io/blog/can-ai-edit-videos): what AI editors genuinely can and cannot do
- [50 AI video editing prompts that work](https://valmera.io/blog/ai-video-editing-prompts)
- [How much does AI video editing cost?](https://valmera.io/blog/how-much-does-ai-video-editing-cost)
- [How to edit talking-head videos](https://valmera.io/blog/how-to-edit-talking-head-videos)
- [AI video editing: the complete guide](https://valmera.io/blog/ai-video-editing-complete-guide)
- [Edit a podcast with AI](https://valmera.io/blog/edit-podcast-with-ai)
- [Turn long videos into clips](https://valmera.io/blog/turn-long-videos-into-clips)

## Comparisons

- [Valmera vs Descript](https://valmera.io/compare/valmera-vs-descript)
- [Valmera vs CapCut](https://valmera.io/compare/valmera-vs-capcut)
- [Valmera vs Veed](https://valmera.io/compare/valmera-vs-veed)
- [Valmera vs Kapwing](https://valmera.io/compare/valmera-vs-kapwing)
- [Valmera vs Captions](https://valmera.io/compare/valmera-vs-captions)
- [Valmera vs Opus Clip](https://valmera.io/compare/valmera-vs-opus-clip)
- [Valmera vs Submagic](https://valmera.io/compare/valmera-vs-submagic)
- [Valmera vs Gling](https://valmera.io/compare/valmera-vs-gling)
- [Valmera vs Wisecut](https://valmera.io/compare/valmera-vs-wisecut)
- [Valmera vs Clipchamp](https://valmera.io/compare/valmera-vs-clipchamp)
- [Valmera vs InVideo](https://valmera.io/compare/valmera-vs-invideo)
- [Valmera vs Vizard](https://valmera.io/compare/valmera-vs-vizard)
- [Valmera vs Runway](https://valmera.io/compare/valmera-vs-runway)
- [Valmera vs Reap](https://valmera.io/compare/valmera-vs-reap): the two video editing MCP servers side by side — 108 tools finishing one video against 10 tools clipping, captioning and dubbing many; Reap wins outright on dubbing (80+ languages), subtitle translation (100+), romanized scripts, batch output, publishing and team seats
- [Best AI video editors](https://valmera.io/alternatives/best-ai-video-editors)
- [Best free AI video editors](https://valmera.io/alternatives/best-free-ai-video-editors)
- [Runway alternatives](https://valmera.io/alternatives/runway-alternatives): the search splits in two, because Runway is a generator — five generative alternatives (Sora, Google Veo, Kling, Pika, Luma) for footage that does not exist yet, and four editing alternatives (Descript, Valmera, CapCut, Gling) for footage that already does
- [Best agentic video editors](https://valmera.io/alternatives/best-agentic-video-editors): the agentic field specifically — Valmera, Mosaic, Descript's Underlord, Goldcast, Cardboard, Reap — with what each agent is actually allowed to do

## Category and MCP

- [Agentic video editor](https://valmera.io/agentic-video-editor): the category page — what an agentic video editor is, and the test that separates it from an AI-assisted one
- [Autonomous video editor](https://valmera.io/autonomous-video-editor): the same category from the autonomy angle — which editing decisions an AI can own (mechanical, craft) and which it should not (intent)
- [Agentic video editing tools](https://valmera.io/agentic-video-editing-tools): the full capability inventory an agentic editor needs, in seven layers, with an honest note on which capabilities are common and which are rare
- [Edit video by chatting with AI](https://valmera.io/edit-video-by-chatting-with-ai): conversational editing with no timeline, shown as a real exchange including the corrections
- [What is MCP for video editing?](https://valmera.io/mcp/what-is-mcp-video-editing): the definitional guide — what the Model Context Protocol is, why editing fits it (the server does the pixels, the model decides), what a video MCP server must expose to be useful, stdio vs streamable HTTP and OAuth 2.1, and the five video MCP servers named
- [Valmera MCP server](https://valmera.io/mcp): connect Valmera to Claude and edit real footage from the conversation
- [MCP setup for every client](https://valmera.io/mcp/setup): the exact config location and literal JSON/command for Claude (web/desktop/mobile), Claude Code, Cursor, VS Code, Cline, Windsurf, Zed, Continue, Goose, LibreChat and the MCP Inspector, plus the `mcp-remote` bridge for stdio-only clients — each marked with what was verified and when, and which auth path (OAuth vs bearer token) it takes
- [Connect Claude to video editing](https://valmera.io/mcp/claude): the step-by-step setup guide for the Claude app and Claude Code, plus the errors worth recognising
- [Video editing in Cursor over MCP](https://valmera.io/mcp/cursor): the mcp.json config for Cursor (global vs project scope, OAuth vs bearer token), a first run, and the screen-recording tools that make it a fit for demo and release video
- [MCP tool reference](https://valmera.io/mcp/tools): every tool the connector exposes, what it does and what it refuses
- [Best video editing MCP servers](https://valmera.io/mcp/video-editing-mcp-servers): every MCP server that edits video — Valmera, reap, OpenClip, Video Jungle, DaVinci Resolve MCP — compared on tool surface, auth and scope, including where the others are the better answer
- [ChatGPT and MCP video editing](https://valmera.io/mcp/chatgpt): ChatGPT's two connector paths — company knowledge/deep research wants read-only `search`+`fetch` tools and cannot drive an editor; developer mode loads any tool surface and permits writes, confirmed in the conversation by default, on the web only. Setup steps, the four places it stops (including an unreconciled contradiction in OpenAI's own docs over which plans get write actions), and what to use instead
- [Claude video editor](https://valmera.io/claude-video-editor): what Claude can and cannot do with video, and how MCP closes the gap
- [Can ChatGPT edit videos?](https://valmera.io/chatgpt-video-editor): no — ChatGPT cannot open footage or render a file. What it genuinely helps with, and what to use instead

## Pricing

- [Plans](https://valmera.io/subscribe): Free $0 (50 one-time credits, spendable, no card, never refills), Creator $30/mo (2,000 credits per billing cycle), Pro $50/mo (4,000), Frontier $100/mo (10,000, on a stronger model for both reasoning and vision). Paid plans open with a 3-day free trial; during a trial the account can spend 10% of the plan's credits, and converting releases the rest. Subscribers also get 20 daily credits on top of their monthly pool. Edits charge credits in proportion to the AI work actually done — there is no flat per-edit price, and simple edits cost the least. Free-plan exports carry a small Valmera mark; every paid plan exports with no watermark at all.

## Optional

- [Full product reference for LLMs](https://valmera.io/llms-full.txt): complete capability, limit and pricing reference in one file
- [About](https://valmera.io/about)
- [Use cases](https://valmera.io/use-cases)