VS

AI Video Summarizers

Video Summarizer review

summarize.martino.im · checked 9 Oct 2026

Open-source video summarizer for URLs and files, with transcript and visual modes.

How it plugs in

—Public API
NoOpen source
YesFree tier

Signal on AI Video Summarizers6.4#4 of 25

This free, open-source project turns online videos and local files into notes for reuse. It works with social, cloud, and local inputs, using either caption or Whisper-based transcription or direct video analysis by a vision-capable model. Access is available through web, CLI, API, Docker, Raycast, and agent interfaces, with an LLM key configured by the user.

Video Summarizer home page

Review

In a line

Video Summarizer converts video URLs or uploaded files into notes you can reuse. It is open source, free, and can run locally or through its web and API interfaces.

What it does

The normal process uses captions when available. Otherwise, Whisper creates a transcript that an LLM then summarizes.

With visual processing, the system sends the video itself—including imagery and audio—to a compatible video model instead of transcribing first. The built-in styles span Q&A, distillation, fact-checking, tutorials, Mermaid diagrams, and essays, while prompts.json supports additional styles.

Podcasts, lectures, conference talks, and social clips can all be turned into notes. A cached transcript can be reused when you change the style, provider, or language, avoiding another transcription.

How it connects

Material can come from YouTube, Instagram, TikTok, Twitter/X, Reddit, Facebook, Google Drive, Dropbox, or local files. Video uploads are supported.

It works with OpenAI-compatible endpoints such as OpenAI, Groq, Gemini, Ollama, OpenRouter, NVIDIA, Perplexity, LiteLLM, and other compatible services. Put at least one provider key in a local .env file.

You can use a Streamlit web app, command-line tool, optional HTTP API, Docker setup, Raycast extension, or agent skill. The project is self-hosted and local-first.

Good fit

This suits people who want notes from long talks, classes, podcasts, or social clips while choosing their own model provider. It can also fit users comparing AI video summarizers who want open-source code and local control.

The optional JEV prefilter can select transcript passages before summarization and include or exclude chosen concepts. Tinypaste can turn a web summary into a public rendered page without requiring an account or token.

Watch for

Plan on Python 3.7 or newer, ffmpeg on your system, and at least one LLM API key. The tool does not include a free trial because the product itself is free.

For visual requests, the documented defaults are 120 seconds and 100 MB per file. MP4, MPEG, MOV, and WEBM are the formats listed in the documentation.

Pages shared through the tool are public. By default, the web app retains only its latest 10 summaries in memory, and refreshing clears that history unless disk persistence is configured. No security certification or compliance standard is stated.

Pricing

Video Summarizer is free and open source. The product has no free-trial period.

Verdict

Choose Video Summarizer if you are comfortable setting up Python, ffmpeg, and an API key and want control over where processing runs. Its broad source support, multiple interfaces, and transcript or visual workflows cover different ways to turn video into notes.

Look elsewhere if you need a hosted service with no setup, private sharing by default, or an included model allowance.

Our take

Video Summarizer is a practical pick for technically confident users who want an open-source workflow and the freedom to select an LLM provider. It handles both online sources and uploaded files, offers several note formats, and can avoid repeat transcription through caching. The trade-off is setup: you must install its prerequisites and bring a provider key. Visual processing has short, size-limited requests, and shared pages are public. It is less suitable for buyers seeking a managed, all-inclusive subscription.

  • Free, open-source software
  • Supports major social, cloud, and local video sources
  • Provides both transcript workflows and direct visual analysis
  • Works with many OpenAI-compatible providers
  • Web, CLI, API, Docker, Raycast, and agent interfaces
  • Requires Python, ffmpeg, and your own LLM API key
  • Visual requests are limited by defaults of 120 seconds and 100 MB
  • Pages shared through the tool are public
  • Web history disappears on refresh unless persistence is enabled
  • No security certification or compliance standard is stated
  • Pick it ifPeople who want local, open-source video notes; Users with their own LLM provider key; Summaries of talks, podcasts, and social videos
  • Skip it ifYou want a hosted service with no technical setup; You need private-by-default shared summaries; You want model usage included in the product price

Pricing

The software is free and has no trial period. An LLM API key is required for configuration.

Questions

Is Video Summarizer free?

Yes. It is open-source software offered at no cost, with no free-trial period.

Which video sources work?

Supported inputs include YouTube, Instagram, TikTok, Twitter/X, Reddit, Facebook, Google Drive, Dropbox, and local files. You can also upload video files.

Do I need an API key?

Yes. Add at least one LLM API key to a local .env file. Examples of supported providers are OpenAI, Groq, Gemini, Ollama, OpenRouter, NVIDIA, and Perplexity.

Does it transcribe every video?

Not always. The standard workflow uses captions or Whisper transcription. Visual processing sends video directly to a vision-capable model without transcribing first.

How long can visual requests be?

The documented visual defaults allow 120 seconds per request and 100 MB per file. MP4, MPEG, MOV, and WEBM are listed formats.

Are shared summaries private?

No. Tinypaste-published summaries appear as public rendered pages.

Can I rerun a video with another style?

Yes. A transcript cache lets you reuse the same transcript across styles, providers, or languages, and disk persistence is available.