Straight answers, short and honest. If yours is not here, email support (at) osinix.com.
Your key arrived by email, so check your inbox first. If it is gone, email licensing (at) osinix.com from the address you purchased with and we will send it again.
Yes, at any time. Payments are handled by Polar, our merchant of record; use the link in your Polar receipt email to manage or cancel, or email support (at) osinix.com and we will take care of it. Your engine keeps running until the paid period ends.
Change your plan in your Polar customer portal, linked from your Polar receipt email. Polar handles the price difference automatically. Once the change lands, we send you the new tier's download and a new license key. If anything is unclear, email licensing (at) osinix.com and we will handle it with you.
Yes. Run the engine with --deactivate on the current machine, then activate on the new one. If you no longer have access to the old machine, email licensing (at) osinix.com and we will free it. Details are in the Licensing Guide.
The quickest route is a single account at OpenRouter or HuggingFace: one key gives you access to many models from many vendors. And local models need no key at all.
Put the OpenRouter model slug in the model field, for example deepseek/deepseek-r1 or x-ai/grok-4.3. You will find the slug on each model's page at openrouter.ai.
Set the endpoint to https://generativelanguage.googleapis.com and enter your key. MCS detects the Gemini protocol from the endpoint automatically. Example model: gemini-2.5-pro.
No. MCS is one self contained download: unpack it, run the installer, add your license key, and it runs. Cloud models need only their API keys. Local models work through whatever you already run, such as Ollama or LM Studio. MCS installs no libraries, no runtimes, nothing extra.
Choose Custom as the provider and point the endpoint at your provider's OpenAI compatible API. That is also how local model servers like Ollama, LM Studio, and llama.cpp connect.
Yes. Go to the free trial page or email freetrial (at) osinix.com, and we will send you a trial license key and a download link.
Not today. MCS runs on Linux and macOS.
No. The engine runs natively on your machine and needs no GPU of its own. Cloud models run on their provider's hardware. Only if you run local models does your machine need what those models need.
Yes. Exactly one participant is the moderator, and the engine refuses to run without one. The moderator opens the session, steers the deliberation, and produces the final deliverable.
Yes, a lot. The moderator steers the whole session and, in directed mode, chooses who speaks on every turn. Give the moderator role to your strongest model.
Two participants, one of which is the moderator. The engine requires an agenda, at least two participants, and a moderator whose name matches a participant.
Yes. A fully local panel works, with no cloud account and no API keys if your model server does not require one. You can also mix local and cloud models on the same panel.
No. Your prompts, files, and results travel directly from your machine to the model providers you configure, or stay entirely on your machine with local models. The only thing MCS sends to Osinix is a periodic license check in containing your license ID, a machine ID, and the engine version. No content, ever.
Cloud models need internet, local models do not. Licensing allows thirty days of offline operation between check ins, so a fully local panel keeps working through network outages.
Yes, really. Participants produce real files during the session, a document, a code file, even an image, and MCS hands each file to the other participants so they can read, critique, or build on it. The shipped two_painters demo shows one model painting an image and another repainting over that actual file.
Two things must both be true. The model must support that output, for example an image capable model. And your tier must allow the file type: Basic exchanges text and source code, Pro adds PDF, and Pro Max and Enterprise allow any type, including images and audio.
The SDK is a C API and a Python package that let your own programs run MCS deliberations. It drives an installed MCS Engine, it does not replace it, so the SDK alone is not enough. It is included with Pro Max and Enterprise, and a $15/mo add on for Basic and Pro. See the pricing page.
We welcome it. Any plan with SDK access lets your own software drive MCS on your machines. If you plan to ship MCS to your customers as part of your product, email sales (at) osinix.com first so we can set up the right agreement.
Start with the documentation: the guides cover installation, usage, licensing, the SDK, and every error code. If the answer is not there, email support (at) osinix.com.
Email support (at) osinix.com. Include your MCS version (mcs --version), your platform, and what happened. Session logs live in each session's logs folder and help a lot.
Email sales (at) osinix.com. Enterprise adds up to 16 participants, unlimited file exchange, server mode, and seat licensing managed centrally.
That is Claire. She presents Osinix MCS, and she appears in our banners and videos.