Resources · Open-Source AI

Open-Source AI Tracker

Open-source AI intelligence filtered for practical builders: models, licences, deployment, hardware and commercial potential.

HIGH PRIORITY · VIDEO

LongCat-Video

Meituan's 13.6B-parameter open video-generation model, covering text-to-video, image-to-video, video continuation and longer-form generation.

Licence

MIT

Open weights

Yes

Self-hosting

Yes

API / hosted

fal.ai · Replicate

Input

Text · Image · Video

Hardware

High-end NVIDIA GPU recommended

Why it matters

The important distinction is not simply that LongCat generates video. Its longer-video focus, open licence, self-hosting path and hosted APIs make it a candidate component for commercial content-production stacks.

Stack position

AI Content → Video Generation → Open Model → Self-host / API

Commercial suitability: promising, subject to verification of the specific model and hosting terms before a commercial recommendation.

VISION · OCR

PP-OCRv6

Lightweight OCR infrastructure for extracting text locally from documents, screenshots and imagery.

Research project →

TRANSLATION

Hy-MT2

Multilingual translation models with a strong local/edge deployment angle and Apache 2.0 licensing.

Research project →

MULTIMODAL

North Micro Vision

Compact vision-language infrastructure for document, OCR and visual-analysis workflows.

Model →

IMAGE

SenseNova U1.5

Open image-generation and editing infrastructure designed for local workflows.

Research project →