MR BEBO / local ai

Intelligence, closer to you.

Real products, clear pipelines and honest answers about what runs on-device.

WHY LOCAL?

Bring the workflow closer.

The lab follows actual products: what runs on-device, how steps connect, and where network access is still needed. Local-first does not mean every feature is offline.

Model versions, licenses and hardware results are published only when supported by product evidence or repeatable tests.

A WORKFLOW, NOT A MODEL LIST

An idea, all the way to export.

A conceptual view of a creative pipeline. The exact steps and online options depend on the product.

01Idea & script
02Local inference
03Voice & assets
04Edit & export

Inside the products

PUBLISHED PRODUCT

Movyra

Local scene planning and script generation

Local rendering and voices for supported workflows

Product source
CROSS-LINKED PRODUCT

Kinetra

Local-first 2D animation studio for Mac

Storyboard, artwork, motion, narration, lip sync

Product source
PUBLISHED PRODUCT

Auvio

Local transcription

Local translation

Local voice generation

Dubbing and timeline synchronisation

Product source

Movyra hardware guidance

PUBLISHED GUIDANCE / 01

minimum

macOS 26+, Apple M1, 8 GB unified, 20 GB SSD

PUBLISHED GUIDANCE / 02

recommended

macOS 26+, Apple M2 Pro/M3/M4, 16 GB unified, 50 GB SSD

PUBLISHED GUIDANCE / 03

high end

Apple M3 Max/M4 Max/Ultra, >=36 GB unified, 100 GB SSD

Movyra product-level guidance from the supplied research dated November 2, 2026. These are not model-specific minimums or measured benchmarks. Linux is listed as coming soon. Read the official requirements ↗

Model catalog

No model manifests have been approved for publication yet.

Measured benchmarks

Device test results will appear when reproducible measurements are available.

LET’S CONNECT

Have something
in mind?✳

An interesting idea, a collaboration, or simply a conversation. I’d love to hear it.

Let’s make it happen