models
15+ · any provider
Everything you expect from the AI assistant you already use: chat, agents, documents, code, and search. Plus free model selection and no vendor lock-in, because the core is open and the data is yours.
The Act phases in through 2027; general-purpose AI duties begin in Aug 2025.
Bring any model, even your own. Pick from 15+ across every major provider, or let the auto-router choose the best one for each task.
GPT, Gemini, Claude, Grok, Mistral, DeepSeek, Qwen, and more.
Match the model to the task, or pick per message.
Run open weights or your own self-hosted models, because the core is open.
A markdown knowledge base with [[wikilinks]] and a cross-resource graph that searches everything you've written, uploaded, or said.
Author links in-text; the graph indexes them on save.
pgvector over pages, files, drafts, and memories.
Local, user, and agent scopes the agent can recall.
Chat is the surface. Underneath: autonomous agents, a knowledge brain, a sandbox, and your whole team, sharing one context.
Chat, brain, agents, files, your team. One context that never loses your place, and never leaves your control.
Run Magus on our EU cloud with simple, transparent pricing, and export your data whenever you choose. Or self-host the open core on your own infrastructure, at no extra fees.
Run Magus on our managed cloud, hosted in the EU. Nothing to set up, and always up to date.
A flat base fee plus pay-as-you-go usage. No hidden costs and no surprise bills, so you always know what you pay.
Your data stays yours. Export everything whenever you choose and leave with no friction.
Prefer your own infrastructure? Run the open core yourself at no extra cost. Your hardware, your rules.
A small Swiss team, building with the community. The core is open source on GitHub and the conversation lives on Discord. Come shape it with us.