The AI Video Stack Has Overtaken Classic Editors: 10 Open-Source Tools Ranked by GitHub Stars
The 10 Hottest Open-Source Video Production & Editing Projects on GitHub in 2026: AI Clip Generation, Code-to-Video, Lossless Editing — How to Choose (with Screenshots)
In 2026, the landscape of open-source video tools has undergone an obvious generational shift.
Previously, the main players in this field were classic tools like FFmpeg and Shotcut; now, the fastest-growing projects on GitHub are AI video agents that "take a sentence and deliver a finished clip automatically." At the same time, programmatic solutions that write video using React/Python are also eating into some of the scenarios traditionally served by professional editing software.
This article does three things:
- Uses the GitHub Search API with multiple keywords (
video editor,video editing,ai video,ffmpeg,视频剪辑,短视频), sorts by stars, and keeps only projects directly related to "making and editing video" that are still actively maintained; - Opens each repository page individually to verify data and take screenshots; all star counts are as of November 21, 2026;
- Provides selection recommendations and boundary notes along three routes: "AI automation / code-programmatic / desktop GUI."
First, an important clarification: stars measure popularity, not quality. Half the projects on this list were created in 2026, so their growth is very fast. Whether they're worth betting on depends on issue response speed and release cadence, which will be noted for each project below.

Quick List
| Rank | Repository | Stars | Language | License | Route | One-line positioning |
|---|---|---|---|---|---|---|
| 1 | harry0703/MoneyPrinterTurbo | 144.8k | Python | MIT | AI automation | Enter a topic, one-click generate vertical/horizontal short videos |
| 2 | 3b1b/manim | 96.7k | Python | MIT | Code-programmatic | The same math animation engine used by 3Blue1Brown |
| 3 | heygen-com/hyperframes | 87.5k | TypeScript | Apache-2.0 | AI automation | Editable AI video, HTML animations converted to MP4 |
| 4 | OpenMontageInc/OpenMontage | 82.0k | Python | AGPL-3.0 | AI automation | Agent-based full-pipeline video production system |
| 5 | remotion-dev/remotion | 74.6k | TypeScript | Custom license | Code-programmatic | Write video in React, engineered batch rendering |
| 6 | mifi/lossless-cut | 46.3k | TypeScript | GPL-2.0 | Desktop GUI | Swiss Army knife for lossless video cutting |
| 7 | ATH-MaaS/Pixelle-Video | 30.4k | Python | Apache-2.0 | AI automation | Alibaba AIDC's fully automated short-video engine |
| 8 | HBAI-Ltd/Toonflow-app | 22.4k | TypeScript | MIT | AI automation | AI short drama and comic creation workbench |
| 9 | motion-canvas/motion-canvas | 20.1k | TypeScript | MIT | Code-programmatic | Use generator functions to make explainer-style animated videos |
| 10 | Huanshere/VideoLingo | 19.3k | Python | Apache-2.0 | AI post-production | Video translation, subtitle alignment, and voice cloning |
These 10 projects are not mutually exclusive; they sit at different stages of the video production pipeline:

I. AI Automation Faction: Give a sentence, get a finished video
This is the most crowded and fastest-growing faction on the 2026 list, taking 6 of the 10 slots. Common traits: LLM writes the script, models generate assets, TTS does voiceover, and FFmpeg handles the final composite export.
1. MoneyPrinterTurbo: 144.8k stars, short-video account growth assembly line

- What it is: Python, MIT license, created March 2024, currently 144k stars and 23k forks. Last code push on November 20, 2026, very actively maintained.
- What it can do: Enter a topic or keyword, it automatically runs the full pipeline of "LLM writes script → matches assets → TTS voiceover → adds subtitles → composes final video," supporting vertical 9:16 and horizontal 16:9, with both WebUI and API usage. The screenshot shows its official Chinese/English case library.
- Who it's for: Teams doing knowledge-based talking-head videos or news roundups that need to batch-test topics.
- Boundaries: Assets mostly come from third-party stock libraries, so copyright must be reviewed yourself; content generated from the same topic is highly homogeneous, so it's only a first-draft pipeline, not a replacement for account personality; running it requires your own LLM and TTS API keys.
2. HyperFrames: 87.5k stars, HeyGen's official open-source editable AI video

- What it is: Open-sourced by digital-human company HeyGen in March 2026, Apache-2.0, TypeScript, reached 87k stars in just 8 months.
- What it can do: Core idea is converting HTML, CSS, and GSAP animations into deterministic MP4 video — meaning every AI-generated frame is directly editable like a webpage, not a black-box output. Provides desktop app and CLI, natively supports MCP, can be driven by coding agents like Claude Code and Cursor. The screenshot shows its built-in animation framework templates.
- Who it's for: Engineering teams needing batch production of motion-graphics packaging, ad templates, and marketing shorts.
- Boundaries: Young project, ecosystem and docs still changing rapidly; strong at motion graphics and template-based production, not suitable for fine editing of real-person footage.
3. OpenMontage: 82.0k stars, a 12-module Agent video factory

- What it is: Python project created March 2026, claiming to be the first open-source agentic video production system, 82k stars and 11k forks.
- What it can do: Splits video production into 12 product modules, integrates external services like OpenAI, Flux, and ElevenLabs, renders on Remotion under the hood, and orchestrates from script and storyboard to voiceover and final output via agents.
- Who it's for: Developers wanting to study multi-agent collaboration for content production, or teams needing customized video pipelines.
- Boundaries: License is AGPL-3.0, which has copyleft requirements that may be problematic for commercial integration; closed-source products must undergo legal review before integration. The entire chain depends on multiple external APIs, so quota bills are non-trivial.
4. Pixelle-Video: 30.4k stars, a locally runnable fully automated short-video engine
![]()
- What it is: Open-sourced by Alibaba's AIDC team, created November 2025, Apache-2.0, built on ComfyUI.
- What it can do: Enter a topic, it automatically completes script, images, TTS, subtitles, and rendering, with a web interface. Workflows can run on local GPUs, so data doesn't leave your domain.
- Who it's for: Teams with data compliance requirements that want fully private deployment.
- Boundaries: Watch maintenance cadence — the last code push was on June 14, 2026. Before adopting, check issue response and confirm compatibility of the required ComfyUI nodes.
5. Toonflow: 22.4k stars, a vertical workbench for AI short dramas and comics

- What it is: Created January 2026, MIT license, TypeScript + ElectroBun desktop, current version v2.0.11, supports 22 interface languages.
- What it can do: Designed for short-drama and comic scenarios, integrating script creation, intelligent storyboarding, character and scene asset management, text-to-image, image-to-video, with an infinite canvas and node-based editor to bring the whole creation pipeline into one app.
- Who it's for: Studios working on short dramas, comics, or interactive narrative content.
- Boundaries: Highly vertical scenario; using it for talking-head product promos is overkill. Node-based workflow has a learning curve.
6. VideoLingo: 19.3k stars, an all-in-one subtitle translation and dubbing solution for video going overseas

- What it is: Created August 2024, Apache-2.0, Python, still actively updated in November 2026.
- What it can do: One command handles speech recognition, subtitle segmentation, translation, alignment, dubbing, and even voice cloning, with a focus on Netflix-level subtitle sentence breaks. Provides Streamlit UI, HTTP API (port 8501), and Docker deployment. Apple Silicon uses MLX acceleration, NVIDIA GPUs use CUDA.
- Who it's for: Creators moving Chinese videos to overseas platforms like YouTube and TikTok.
- Boundaries: It does post-production localization, not content generation from scratch. Translation and lip-sync quality depend on source audio quality and chosen models; professional finished output still needs human review.
II. Code-Programmatic Faction: Video as code, versionable and batchable
This faction doesn't chase trends but has the longest lifecycle. Core value is turning video into an engineering artifact: go into Git, be code-reviewed, and render ten thousand variants at once.
7. Manim: 96.7k stars, the de facto standard for explanatory animation

- What it is: Python animation engine open-sourced by 3Blue1Brown in 2015, MIT license, 97k stars. Note there are two branches: the original author's version and the community-maintained version (ManimCommunity). New users are advised to start with the community edition.
- What it can do: Precisely control every movement of formulas, shapes, and camera using code, producing explainer videos for mathematical proofs and algorithm principles.
- Who it's for: Creators doing science communication, teaching, or technical evangelism.
- Boundaries: Steep learning curve; debugging an animation can be slower than editing a clip. It only generates animations, cannot process real footage.
8. Remotion: 74.6k stars, the leading way to write video with React

- What it is: TypeScript project created in 2020, current v4.0.545, ~16 million npm monthly downloads, the most commercially mature in this category.
- What it can do: Describe visuals with React components and drive animations with a timeline, plus a Remotion Studio with preview (left side of screenshot) for code-and-visual dual-mode editing. Supports cloud batch rendering on Lambda, suitable for personalized, per-user videos.
- Who it's for: Front-end teams making product demos, data reports, and batch marketing assets.
- Boundaries: Uses a custom commercial license rather than a standard open-source license. If employee count or usage scale exceeds thresholds, a company license is required. Read the License page before commercial use.
9. Motion Canvas: 20.1k stars, an animation library built for technical explanations

- What it is: Created in 2022, MIT license, TypeScript.
- What it can do: Use generator functions to orchestrate animations with a live preview editor, so code logic and visual rhythm naturally stay in sync. Particularly suited for explaining abstract concepts like algorithms, concurrency, and state machines.
- Who it's for: Programmers who don't want to learn editing software but want high-quality technical explainer videos.
- Boundaries: Ecosystem size and template count are clearly smaller than Remotion. It's positioned as animation generation, not general editing; real-footage handling is weak.
III. Desktop GUI Faction: Offline, deterministic, lossless
No matter how hot AI gets, there's a class of needs it can't replace: no network, deterministic results, zero damage to original footage.
10. LosslessCut: 46.3k stars, the lossless-cutting GUI for FFmpeg

- What it is: Veteran project created in 2016, GPL-2.0, Electron-based, current v3.70.2, cross-platform macOS/Windows/Linux.
- What it can do: Essentially a GUI wrapper for FFmpeg that cuts, merges, and extracts audio/video streams at keyframes without re-encoding, so it's extremely fast and has zero quality loss. The screenshot shows its segment export list and multi-track timeline. Also provides CLI and HTTP API for automation. Ideal for quickly trimming dozens of GB of footage from GoPro, drones, or cameras.
- Who it's for: Anyone needing to quickly discard bad takes, split long recordings, or extract audio tracks.
- Boundaries: "Lossless" also means no transitions, color grading, or subtitle effects; for those, you need a professional NLE.
Selection Advice: Choose the route first, then the project

Direct recommendations by scenario:
| Your scenario | First choice | Backup |
|---|---|---|
| Batch knowledge talking-head, matrix account growth | MoneyPrinterTurbo | Pixelle-Video (when you need privatization) |
| Motion-graphics packaging, template-based ad batch output | HyperFrames | Remotion |
| Short drama, comic, interactive narrative | Toonflow | OpenMontage |
| Video overseas translation and dubbing | VideoLingo | OpenCreator |
| Math/algorithm/technical principle explanation | Manim | Motion Canvas |
| Real-footage rough cut, lossless cutting | LosslessCut | Shotcut |
Finally, three easily overlooked reminders:
- License risk: OpenMontage is AGPL-3.0, LosslessCut is GPL-2.0, both unfriendly to closed-source commercial use; Remotion has a custom commercial license, check the terms before company use. MIT and Apache-2.0 projects are far more permissive.
- Real cost: The AI automation faction saves labor but shifts costs to API calls and GPU rental. Before running a full local pipeline, don't plan the budget as if it were a production environment.
- Popularity doesn't equal maturity: HyperFrames, OpenMontage, and Toonflow are all new repositories from 2026 with astonishing star growth. For production selection, observe three months of issue response and release frequency, or pin to a fixed version, to avoid rework from mainline changes.
Honorable mentions that didn't make the list but are worth saving: the infrastructure base FFmpeg (67.7k stars, the common foundation for almost all projects), professional NLE Shotcut (15.7k), Python video library MoviePy (15.2k), video translation tool OpenCreator (14.4k), Alibaba subtitle editor FunClip (6.6k), and auto-editor (6.0k) which automatically cuts silent segments.
References
- harry0703/MoneyPrinterTurbo
- 3b1b/manim
- heygen-com/hyperframes
- OpenMontageInc/OpenMontage
- remotion-dev/remotion
- mifi/lossless-cut
- ATH-MaaS/Pixelle-Video
- HBAI-Ltd/Toonflow-app
- motion-canvas/motion-canvas
- Huanshere/VideoLingo
Data source: GitHub REST Search API, search date 2026-11-21; repository screenshots taken by visiting GitHub repository pages; demo visuals copyright belong to their respective owners.