Model
Open-source self-host = always free. Multica Cloud is currently in waitlist state per the docs ("No. 01 Multica Cloud — Managed backend… Waitlist"). Pricing for the cloud tier is not disclosed yet.
What's free
- The full open-source platform — Apache or MIT class license (license not surfaced in captured content; standard for the OSS positioning).
- All features: Agents as Teammates, Skills, Unified Runtimes, multi-workspace, full task lifecycle, WebSocket streaming, CLI, Desktop app.
- "As many [agents] as your hardware supports. Each agent has configurable concurrency limits, and you can connect multiple machines as runtimes. There are no artificial caps in the open source version."
What costs money
The user pays for:
- Their own LLM API costs — agent execution happens on the user's machine via their installed CLI (Claude Code → Anthropic, Codex → OpenAI, etc.). Multica brokers, doesn't bill.
- Their own compute — agents run on user hardware (local daemon) or user-provisioned cloud runtimes.
- Future cloud tier — amount and structure unknown.
Versus the rest of the vault
| Axis | Multica | Replit | Anything | Fly.io | Hercules |
|---|---|---|---|---|---|
| Free tier | ✅ All features (self-host) | ✅ Starter | ✅ Free credits | 🟡 (scope unclear) | ✅ "Generous free hosting" |
| Paid tier shape | ❓ Cloud = TBD | Tiered + bundle | Tiered + Max add-on | Pure metered | Tiered (specifics not captured) |
| Pricing transparency | All-free for OSS | Public | Public | Calculator-only | ❌ Pricing page didn't render |
| LLM cost | ✅ User pays directly | Bundled in plan | Bundled in plan | N/A | Bundled in plan |
💭
This is the clearest pricing differentiator in the vault — Multica is the only competitor where the user pays the LLM provider directly. Builders bundle inference into credits and absorb the margin. Multica is happy to broker without owning the AI cost. For agent-economy CreateOS positioning, this is interesting: Multica leaves the door open for its agents to use any router — including CreateOS's router.nodeops.network.
Sources
- multica.ai/docs (Welcome page)
- github.com/multica-ai/multica (README + FAQ)