MaystroRequest access

Local-first AI creative suite for Windows

The AI studio that lives on your GPU.

Maystro brings image generation, cinematic video with sound, character consistency, and LoRA training into one desktop studio. No cloud queue, no uploads — your work never leaves your machine.

See it in action

Real product, real output — every frame below was generated locally on a single consumer GPU.

Maystro Storyboard workspace with scenes, shots and completed takes
Storyboard: organize scenes, shots and takes — cinematic video with synchronized sound.
Maystro voice assistant home screen
One assistant across every studio — type it or say it.
Maystro model and preset management page with per-slot file report
Transparent model management: every preset shows exactly which files it loads, and any slot can be rewired.

Four studios, one workspace

Everything from a first sketch to a finished shot happens in the same app, on the same machine.

Generate, refine, repeat

Image Studio

State-of-the-art image generation on an editable canvas. Upscale, inpaint, and edit with instruction-based tools — including camera-angle control over a finished image.

  • Canvas editing
  • Upscaling
  • Instruction edits
  • Camera angle

Storyboard-driven filmmaking

Video Studio

Produce video the way a set works: projects, scenes, shoots, and takes. Generate cinematic shots with synchronized audio, then iterate take by take until it lands.

  • Scenes & takes
  • Audio + video
  • Image-to-video
  • Preset control

Your data, your style

Training Studio

Fine-tune LoRAs on your own material with dataset management, hyperparameter control, and live training progress — all running on your GPU.

  • Dataset tools
  • LoRA training
  • Live progress

Identity that holds

Character Studio

Design a character once and keep them recognizable across images and shots, with a dedicated character editor and angle builder.

  • Character editor
  • Angle builder
  • Cross-shot identity

Local-first is the whole point

Maystro is built around one rule: generation happens on your hardware, and only there.

01

100% local generation

Every pixel and every frame is rendered on your machine. The only network traffic is license activation and downloading updates or model weights — never your prompts, images, or footage.

02

Big models on one GPU

Maystro's streaming engine moves model weights between RAM and VRAM just in time, so 20B-class video models run on a single high-end consumer GPU instead of a server rack.

03

Bring your own models

Point Maystro at the model folders you already have, see exactly which file fills which slot, and derive custom presets from any base — no re-downloading, no lock-in.

One plan, no meters

Your GPU does the work, so there are no credits, no render fees, and no per-seat tiers.

Maystro Studio

$299per year

  • All four studios — image, video, training, characters
  • Unlimited generation on your own hardware
  • A year of updates and new model support
  • Perpetual fallback — stop renewing and your installed version keeps working
  • One-machine license with a generous offline grace period
Request access

Maystro is in early access. Request access and we'll set you up personally — payment by invoice, license key by email.

Frequently asked questions

What do I need to run Maystro?

A Windows 10/11 PC with a modern NVIDIA RTX GPU. Image work is comfortable on mid-range cards; video generation benefits from 24 GB of VRAM and 32–64 GB of system RAM. Exact requirements depend on the models you choose.

Does anything leave my machine?

Not your work. Generation runs entirely locally. Maystro talks to our servers only to activate your license and to download updates or starter model weights — and it keeps working offline for extended periods between checks.

What happens if I stop renewing?

The last version you installed keeps working forever — that's the perpetual fallback. You stop receiving updates and new model support until you renew.

Which models does it ship with?

Maystro downloads a curated starter set of open-weight models at install time, and it's built for bring-your-own-model workflows: add your own checkpoints and LoRAs, and assign them per preset slot. Model weights come under their own licenses.

How many machines can I use?

One machine per license (node-locked). You can deactivate a machine and move your license to a new one.

How do I get it?

We're in early access. Send a request through the form, we confirm payment by invoice, and you receive your license key and installer by email — usually within a day.