Find
Scout 5 platforms in parallel. Every result gets a 0–100 virality score and an outlier ratio against its own channel's median, so a 20× breakout is visible before you watch it.
Trending FinderScout. Clip. Generate.
ViralMint is an open-source desktop app for short-form video: it finds what's trending across 5 platforms, clips the moments worth posting from 1,800+ sources, and generates finished shorts with AI voice, captions and music. Pay per video, no subscription.
Free download · AGPL-3.0 · macOS, Windows, Linux · no card to start
- ★ 121 on GitHub
- 1,800+ sites via yt-dlp
- 5 platforms scouted in parallel
- 17 AI video models, billed per second
- 3 OS macOS · Windows · Linux
In action
One pipeline. Every part of the workflow.
Chat with your content strategist.
Paste a niche or video URL — AI plans, scouts, downloads, transcribes, scripts and assembles. All in one chat thread.
What ViralMint does
Three steps from trend to finished short.
- YouTube
- TikTok
- Douyin
- Bilibili
- +1,800+ more
Clip
Paste a long video from 1,800+ sites. Whisper transcribes it on your machine; AI picks the 30–60s moments worth posting and cuts them with word-by-word captions.
ClipperGenerate
A topic or script becomes a finished short: AI voice, captions, music, stock or AI b-roll. Motion graphics, AI video clips (17 models) and a cloned voice are one click away.
Smart VideoReleased April 2026 under the AGPL-3.0 license. Downloads, Whisper transcription and rendering run on your computer; AI generation runs through your account with no API keys to manage. You download the finished mp4 and post it yourself with the AI-drafted title, description and tags. Compare with other tools · FAQ
Features
Everything to find, study, and ship viral videos.
From trending discovery to a finished mp4 with AI titles, descriptions and tags.
- AGPL-3.0
- No API keys
- Runs on your machine
- No watermark
Trending Video Finder
Five platforms scouted in parallel. 0–100 virality score per result.
Outlier Detection
Each video compared to its channel's own median. 3× / 5× / 10× / 20× breakouts surfaced.
Whisper + AI Analysis
yt-dlp + local Whisper. AI extracts hook, structure, tone and angle in one pass.
Smart Video Pipeline
AI script + Pexels stock + voice + word-by-word captions + auto music, stitched into one mp4.
Clipper
Drop a long video or podcast — AI finds the best 30–60s moments, cuts them, and burns word-by-word captions.
Motion Graphics
AI-authored kinetic typography, stat cards and lower-thirds — rendered to mp4 locally from HTML, no real footage.
Website to Video
Paste any URL — ViralMint screenshots the site, reads the brand, and builds a promo video automatically.
AI Image & Video Clips
Prompt → four image variations → animate the pick into a clip. Auto model pick across Nano Banana, Sora 2 Pro, Veo 3.1, Seedance and Kling; billed per second.
Voice Cloning
Clone a voice from a short clip and narrate in it — runs 100% on your machine, your audio never leaves the device.
Chat Runs Everything
Ask for it in plain language — scout, download, clip, caption, generate. The assistant operates the app and shows you the price before it spends.
Run It From Your Phone
Connect Telegram, WhatsApp, Slack or Discord and drive the same pipeline by message — start a job on the train, get the mp4 when it lands.
MCP Server for Claude & Cursor
Drive the whole pipeline — scout, download, clip, generate, export — from Claude, Cursor or any MCP client in plain language.
Measured on real hardware
Fast enough on the laptop you already own
The heavy work — downloading, Whisper transcription, captioning and the final FFmpeg render — runs on your own machine, not on a rented GPU. So the honest question is how long that actually takes. These are our own measurements on a mid-range laptop (8 GB RAM, CPU only — no GPU required), not projections:
- ~45s to transcribe 10 minutes of audio faster-whisper, int8 quantisation, running on your CPU
- 2-5 min to assemble a 60-second short FFmpeg stitch, voice mix, music bed and burned-in captions
- <3s to scout 5 platforms at once parallel async workers, not sequential requests
- <5ms overhead per MCP tool call loopback HTTP to the local server — no cloud round-trip
No GPU is required and nothing is queued behind other users — the only steps that leave your computer are the AI ones you choose to pay for. Figures are self-reported on the hardware above and will vary with your CPU, the source file and how much else is running.
Pricing
No subscription. Pay only for what you generate.
Prepaid top-ups — small or large. Same balance on desktop + browser. Open-source under AGPL-3.0.
Trial
$0 no card
Try ViralMint with no commitment. Same account upgrades later.
- Daily starter allowance to evaluate the AI features
- Allowance covers any AI feature — chat, image, voice, music
- Edge TTS — 400+ voices at a fraction of a cent
- Same login across the desktop app and the web
Starter
$5 one-time
Small top-up when the daily allowance isn't enough.
- $5 prepaid balance (1:1, no bonus)
- Stacks on top of the daily starter allowance
- Allowance is spent before prepaid balance
- No subscription, no auto-renew
Most Popular
Pro
$30 → $35 balance
Best for regular creators. +$5 bonus credit on top up.
- $35 prepaid balance (+$5 bonus)
- Same per-action prices as Starter
- Per-user $30/day spend cap (configurable)
- Same balance on desktop and browser
Max
$100 → $125 balance
For studios and agencies. +$25 bonus credit on top up.
- $125 prepaid balance (+$25 bonus)
- Same per-action prices as Starter / Pro
- Daily starter allowance still applies on top
- Same balance on desktop and browser
What things cost
Per-action prices, the same on desktop and in the browser
Downloading, Whisper transcription, captioning and reframing run on your machine for a flat 1¢ per job. Everything else below calls a cloud model.
| Action | Price | What is included |
|---|---|---|
| Smart Video short (stock footage) | 2¢ | Script, Pexels b-roll, captions and music assembled on your machine. Voice is billed separately: an Edge voice is 0.5¢, Gemini is below. |
| AI voice-over (Gemini) | 12¢ / 1K chars | A 60-second script is about 900 characters. The 400+ Edge TTS voices are 0.5¢ a generation. |
| Voice clone narration (on-device) | 2.5¢ / 1K chars | Capped at 10¢ per generation. Your reference audio never leaves your machine. |
| Clip extraction | 0.5¢ / clip | Billed per clip produced. The cut runs on your machine; downloading the source is its own row below. |
| Video download | 1¢ / video | 1,800+ sites via yt-dlp, at source quality. |
| AI image | 6¢ / image | Nano Banana, Gemini 3 Pro Image, GPT Image 2, FLUX.2 Klein. |
| AI video clip | 2¢–6¢/s to 10¢–50¢/s | Billed per second by model: budget (Wan, Hailuo) up to flagship (Sora 2 Pro, Veo 3.1). Max 15s per clip. |
| Motion graphics video | 50¢ | A Claude Opus design plan, one AI compose and a local render; each chat message with the AI director is 12¢. Refinements re-bill the compose only. |
| AI music track | 15¢ | Lyria 3 Pro, royalty-free, 12 genres. |
Prices are deducted from your prepaid balance per action; a daily starter allowance covers trying things out. Unused prepaid balance is refundable where the law requires it, and we refund any charge our cloud got wrong — refund terms.
FAQ
Frequently asked questions
Does ViralMint work on a phone?
ViralMint is a desktop app for macOS, Windows and Linux, because downloading, Whisper transcription and rendering run on your own computer. Without a computer, the browser version at viralmint.net/app covers the cloud features (chat, AI image, AI voice, AI music) on the same account. With the desktop app running, you can also drive it by message from Telegram, WhatsApp, Slack or Discord.
Is ViralMint free?
The app is free to download and open source under AGPL-3.0. Using it spends a prepaid balance per action: jobs that run on your own machine, such as a download, a transcription or a caption burn, bill a flat 1¢ each, and cloud AI generation is priced per action in the table above. The open-source build on GitHub runs on your own API keys instead.
How do I install it?
Pick the installer for your system in the download section below, open it, and sign in with your email. There are no API keys to paste: AI features run through your ViralMint account.
What does a finished video cost?
Every action is priced up front in the table above, and the app shows the price before it spends anything. A stock-footage Smart Video short is the cheapest route to a finished video; AI video clips cost more and bill per second by model.
Do I need a GPU or API keys?
No. Transcription and rendering run on an ordinary CPU, and the large AI models run in the cloud through your account, so there are no provider keys to manage.
Models and engines you can use
- Nano Banana
- Lyria 3 Pro
- Gemini 3.1 Flash TTS
- VoxCPM
- Edge TTS
- Whisper
- OpenRouter
- HyperFrames
- FFmpeg
- yt-dlp
- Pexels
- Pixabay
Download
Get ViralMint on your machine
Native installers for macOS, Windows, Linux. Whisper, FFmpeg and yt-dlp run on your machine; AI generation runs through your account.
macOS
Universal — Apple Silicon & Intel
Download .dmg → v2026.10.02Windows
Windows 10/11, 64-bit
Download .msi → v2026.10.02Linux
.tar.xz — Ubuntu, Debian, Fedora
Download .tar.xz → v2026.10.02Or create your free account → (it carries over to the desktop app)