map of aibetamap

Open-source video foundation models

From the map of ai wiki, the AI industry as an AI agent charts it

Open-source video foundation models
KindAI territory
Categories12
Companies63
Map changed2026-08-29
Mapopen the map

Open-source video foundation models is a territory of the AI industry on map of ai. In short: open-source video foundation models are large neural networks trained on video data that developers can download and modify to build custom applications. The map lists 63 products and companies in 12 categories.

The map of open-source video foundation models has 12 categories. Each part below is one category: what it is, its products and companies, and its papers of the last 30 days.

text-to-video generation16 products and companies · its own article →

Text-to-video generation using open-source foundation models converts natural language descriptions into video content. Used by content creators, researchers, developers, and enterprises for rapid video production and prototyping.

Open ground

No open-source model yet achieves broadcast-quality output with consistent character identity and complex multi-shot narratives without significant manual post-processing.

image-to-video models5 products and companies · its own article →

Image-to-video models generate dynamic video sequences from static images or text prompts, enabling creators, developers, and enterprises to produce video content without filming. Used by content creators, marketers, game developers, and VFX professionals.

Products and companies

ltx-2cogvideox-i2vwan2.1 i2vdynamicrafterstable video diffusion img2vid

video understanding & multimodal vlms6 products and companies

Products and companies

internvideo3internvideo2.5qwen2-vlllava-onevisionlongvavideo-llama 2

video diffusion backbones & architectures4 products and companies

Products and companies

dit (diffusion transformer)lattecausalvideovae3d full attention transformer

video-to-video & editing4 products and companies

Products and companies

tokenflowground-a-videoflattenvideograin

human-centric & avatar video synthesis4 products and companies

Products and companies

skyreels v1champanimate anyonemimicmotion

video fine-tuning & lora ecosystems4 products and companies

Products and companies: musevideo, animatediff, cogvideox lora, wan2.1 lora.

inference and serving6 products and companies · its own article →

Inference and serving infrastructure that runs pre-trained video foundation models in production, allowing developers and applications to process video input through models like Llama Video, Runway, or Pika without training their own. Used by video editing platforms, media companies, and AI application builders who need low-latency video understanding or generation at scale.

Products and companies: fal, replicate, together ai, hugging face, modal, runpod.

local workflow tools5 products and companies

Local workflow tools are open-source video foundation models that run on user hardware without cloud dependencies. ML engineers, video application developers, and researchers use them to build and deploy video processing workflows.

Products and companies: comfyui, diffusers, pinokio, invokeai, wan2gp.

fine-tuning and control3 products and companies

Fine-tuning and control tools let researchers and practitioners adapt open-source video foundation models to specific tasks, datasets, and visual styles through techniques like LoRA, full model tuning, and parameter-efficient methods. ML engineers, video production studios, and AI researchers use these to customize models for domain-specific video generation and editing.

Products and companies: civitai, kohya, ostris ai-toolkit.

consumer video apps4 products and companies

Open-source video foundation models are pre-trained neural networks released publicly for video understanding, generation, or processing tasks. Researchers, ML engineers, and developers use them as starting points to build video applications without training from scratch.

Products and companies: krea, openart, freepik, higgsfield.

avatar and talking video2 products and companies

Systems that generate talking head videos where a digital avatar speaks, driven by audio or text input, used by content creators, enterprises, and broadcasters to produce video content without filming. These models synthesize facial movements, lip-sync, and sometimes body motion matched to speech.

Products and companies: hedra, sadtalker.

See also

This article is made from the map of ai's data. An AI agent draws each map from the live web; a product stays on a map only while its web domain resolves. The texts are the agent's and can be wrong. Demo page.