Linum

Type text, dream video.

A two-person research lab training text-to-video models from scratch.

A cute 3‑D animated baby goat with shaggy gray fur, a fluffy white chin tuft, and stubby curved horns perches on a round wooden stool. Warm golden studio lights bounce off its glossy cherry‑red acoustic guitar as it rhythmically strums with a confident hoof, hind legs dangling. Framed family portraits of other barnyard animals line the cream‑colored walls, a leafy potted ficus sits in the back corner, and dust motes drift through the cozy, sun‑speckled room.
A thick golden stream of raw honey pours slowly from a wooden dipper onto a towering stack of fluffy buttermilk pancakes, pooling in the textured surfaces before cascading down the sides in glistening amber rivulets. Soft morning light streams through a nearby window, illuminating steam rising from the warm stack. A pat of butter melts on top, besides fresh blueberries and a dusting of powdered sugar
A wide‑eyed giant panda lounges against a mossy bamboo trunk chewing on a jade-green stalk. Verdant shoots frame the scene while ferns blanket the ground, and tiny flakes of bamboo sprinkle its muzzle. The panda looks surprised, ears perked, as a cicada buzz echoes nearby.
Download model weights →
Linum v2 · 2B params · Apache 2.0
Source code@linuminc

FIELD NOTES

Research, releases, and real-world lessons building generative video.

From our partners · Jun 22, 2026Crusoe
How Linum built a frontier video model on Crusoe ↗︎
Road to Linum v3· Issue 01Research · Aug 26, 2026

Data Filtering for Generative Video Pre-training

Getting Video Models to Learn Better, Faster
22 min read
Tools · Mar 4, 2026

Building a $8/Month GPU-Cluster Monitor

Maintaining our sanity, training a foundation model over the holidays
6 min read
Research · Feb 23, 2026

Better Reconstruction ≠ Better Generation

Learnings from 4 months of Image-Video VAE experiments
20 min read
Read more →

Get Field Notes in your inbox