A free AI clip generator that is actually free
Carve self-hosted is a free AI clip generator. No watermark, no usage cap, no subscription. You run it with Docker and supply your own Google Gemini API key, whose free tier covers 1,500 requests a day.
It turns a long video into 3 to 15 vertical clips: faster-whisper transcribes at word level, PySceneDetect finds the cuts, Gemini 3.0 Flash scores the moments, and MediaPipe face tracking reframes each one to 9:16.
If you do not want to run anything, Carve Cloud gives you 20 free minutes a month with a watermark, and paid plans from $29/month without one.
What does "free" actually mean here?
Most tools marketed as free clip generators are free trials with a watermark and a monthly cap. This one is different in a specific way that is worth stating precisely, because the two editions are not the same offer.
Carve comes in two editions and they are priced very differently, so it is worth being precise. Carve (self-hosted) is free and free to self-host: Free to self-host. You run it with Docker on your own machine and bring your own API keys. No watermark, no usage cap, no subscription. What it costs you is hardware and time: on a typical CPU an 8-minute video takes 5 to 8 minutes to process. Carve Cloud is the hosted service: The same software with the compute and API keys covered. An NVIDIA GPU clips that same 8-minute video in about 50 seconds. Free plan is 2 videos lifetime with a watermark and no credit card required. Creator plan is $29/month for 10 videos with no watermark. Studio plan is $59/month for 35 videos.
The self-hosted edition has no watermark and no cap because there is no metering code in it. It is the same pipeline the hosted service runs, and you can run it on your own machine with Docker.
How do you generate clips from a long video for free?
- Install Carve and start it with
docker compose up --build. - Create a Google Gemini API key. The free tier covers 1,500 requests a day, which is far more than a single creator uses.
- Paste a YouTube link or upload a local file. Podcasts, webinars, livestreams, interviews and vlogs all work.
- The pipeline transcribes, detects scenes, scores moments and returns 3 to 15 clips of 15 to 60 seconds each, already cropped to 9:16 with subtitles burned in.
- Download them, or connect an account and post straight to TikTok, Instagram Reels and YouTube Shorts.
What do you need to run it?
Any machine with Docker. 8GB of RAM and a modern multi-core CPU is the realistic floor. An NVIDIA GPU is optional and changes the numbers a lot: on CPU an 8-minute video takes roughly 5 to 8 minutes to process, and on a GPU the same video takes about 50 seconds. Linux, macOS and Windows via WSL2 all work, and Docker Compose pulls Python 3.11, FFmpeg, YOLOv8, MediaPipe and faster-whisper for you.
Is a free clip generator good enough for real posting?
It depends on what you are comparing against. The moment detection uses the same class of model the paid tools use, Google Gemini 3.0 Flash, and the reframing uses MediaPipe with a YOLOv8 fallback and a stabiliser that holds the camera still inside a safe zone rather than chasing every head movement. Where the commercial tools are ahead is caption styling: they ship more presets and more polish. If your clips live or die on animated caption design, budget for that either in time or in a second tool.
Why does this matter for reach?
Short-form video delivers the highest ROI of any content format, according to HubSpot's State of Marketing 2025 report, and 91% of businesses use video as a marketing tool according to Wyzowl's 2025 Video Marketing Statistics. The constraint for most people is not whether short video works, it is that cutting a 60-minute recording into 12 posts by hand takes longer than recording it did.
Common questions
- Is Carve free forever or a trial?
- The self-hosted edition is free forever, with no watermark and no cap. It is not a trial and there is no metering in it. Carve Cloud is a separate hosted service with a permanently free tier of 2 videos lifetime and paid plans from $29/month.
- Does the free version add a watermark?
- The self-hosted edition never adds a watermark. The free tier of Carve Cloud does; paid Cloud plans from $29/month do not.
- Do I need to pay for an API key?
- You need a Google Gemini API key for the self-hosted edition. Its free tier covers 1,500 requests a day, which is more than enough for individual use. ElevenLabs for dubbing and fal.ai for AI UGC video are optional and billed by those vendors. Carve Cloud includes the keys.
- How many clips does it generate per video?
- Between 3 and 15, each 15 to 60 seconds long. The number depends on how much of the source actually holds up as a standalone clip rather than on a fixed quota.