Skip to content

Docs

Get up and running

Install the app, connect a model provider or use official credits, and keep work local when it matters.

What you need

The Windows build runs on 64-bit Windows 10 (21H2+) and Windows 11.

  • 8 GB RAM minimum; 16 GB or more if you plan to run local models or generate video.
  • A GPU with 6 GB+ VRAM is recommended for local image and video models.
  • An account is optional: local models work without signing in.

Install on Windows

Download the installer from the download page and run it. The installer is per-user and does not require administrator rights.

Verify the SHA-512 hash before running it if you want to be sure the file was not modified in transit.

certutil -hashfile starolestudio-0.1.0-setup.exe SHA512

Add a provider key (BYOK)

StaroleStudio does not require our account to talk to cloud models. Add your own key and calls go straight from your machine to the provider.

  • Open Settings → Providers and choose a provider (OpenAI, Anthropic, Google, DeepSeek, …).
  • Paste your API key. It is stored encrypted in StaroleStudio Vault and never synced in plain text.
  • Pick a default model, or let the router choose based on the task.

Use official models and credits

If you do not have provider keys, buy an official plan instead. Requests then go through the StaroleStudio gateway, which reserves and settles credits on the server.

  • Sign in and open Pricing to choose a plan.
  • Credits are shared across chat, image and video calls.
  • Every request carries an idempotent request id, so a retry is never charged twice.

Run local models

Local models keep prompts and documents on your machine. StaroleStudio connects to a local runtime instead of a cloud API.

  • Install Ollama or llama.cpp, then start the local server.
  • In Settings → Local, point the app at the local endpoint and refresh the model list.
  • For image and video, point the app at your ComfyUI instance.

Troubleshooting

Start with these checks before filing a report.

  • A model is missing: refresh the provider list; some providers require extra model permissions on your key.
  • 401 errors: the access token expired; the app refreshes it automatically. If refresh fails, sign in again.
  • 402 errors: your credit balance is too low for the selected model.
  • Streaming stops mid-answer: check your network or proxy; the app keeps the partial message so you can retry.

FAQ

Where are my keys stored? In an encrypted local vault; they are never sent to StaroleStudio servers.

Can I use my own local models offline? Yes. Local models and local RAG work without an account.

Do official plans replace BYOK? No. They are an optional path for people who do not want to manage provider keys.