whereismyai
Bernini
Bernini is an open-source video generation and editing framework combining an MLLM semantic planner and DiT renderer.
Bernini is an open-source framework for video generation and editing developed by ByteDance. It combines a multimodal large language model based semantic planner with a diffusion transformer based renderer to decompose complex instructions and plan semantic changes before rendering. The repository provides two model families, the full Bernini pipeline and the renderer-only Bernini-R model, supporting tasks such as text-to-image, image-to-image, text-to-video, video-to-video, and more. It runs locally using Python and requires a CUDA GPU, with recommended support for Hopper GPUs and FlashAttention. Users can run inference via command-line scripts using case files or launch a Gradio demo interface.
Bernini at a glance
- Best for: Software engineer, Social media creator
- Runs on: open source, CLI
- Pricing: Free
- Released: May 2026
- Category: AI video tool
- Website: github.com/bytedance/Bernini
Key features
- Text-to-video and video-to-video generation
- Semantic video editing and style transfer
- Multimodal large language model semantic planner
- Diffusion transformer based video renderer
- Prompt enhancer via OpenAI-compatible endpoints
- Gradio demo user interface
- Training and inference code release
Use it for
- Generate videos from text instructions
- Edit existing videos using text prompts
- Remove subtitles or watermarks from videos
- Perform style transfer on video inputs
Alternatives to Bernini
- Fliki — Generates narrated videos from text or prompts using 2,000+ AI voices.
- Wan — Provides native 30-second video generation and pixel-perfect consistency.
- FreeVideo — Runs MiniMax H3 locally on consumer GPUs with as little as 8 GB of VRAM.
- Hypit — Clone viral videos and generate 100 variants with a single command.
- OpenStoryline — It enables conversational video editing, script generation, and workflow automation.
- TurboDiffusion — Accelerates video diffusion generation by up to 200 times on single GPUs.
- ATAS — Maintains persistent spatial and event memory across shots.
- MAGI-1 — Generates synchronized audio and visuals with high-fidelity camera control.
- Kiteva — Generates complete motion-graphics films from simple ideas with local narration and code-built music.
- Pireel — It combines chat-based AI editing, scene-aware tools, and agent connectivity.
- Toonflow — Automates the entire video creation workflow with multi-agent collaboration and character consistency.
- FastVideo — Provides end-to-end post-training and real-time inference for accelerated video generation.
All Bernini alternatives