Pricing
No license fees for the software. You only pay for setup, integration and optional maintenance. AIMOS is and will remain Open Source.
one-time fee, fully credited upon project commissioning
One-Time Services
From hardware consulting to production-ready operation.
| Service | Description | Price (net) |
|---|---|---|
| Hardware Consulting | GPU recommendation, server specification, procurement support | 600 – 1,800 EUR |
| SovereignNode Installation | Operating system, AIMOS, LLM runtime (SGLang), PostgreSQL, SSL, VPN (Tailscale), firewall | 2,500 – 5,000 EUR |
| First Agent (custom) | System prompt, personality, skills, connectors, testing with real data | 1,800 – 3,600 EUR |
| Each Additional Agent | Configuration, skills, connectors, testing | 1,000 – 1,800 EUR |
| ERP / CRM Integration | SQL mapping, API connection, testing with production data (per system) | 1,200 – 3,600 EUR |
| CAD / Document Integration | AutoCAD, Tekla, PDF, Office formats — reading, analyzing, bills of materials | 1,200 – 3,600 EUR |
| Agent Migration | Migrate agents from our server to your SovereignNode (including memory) | 600 – 1,200 EUR |
Typical first project: SovereignNode + 2 agents + 1 ERP integration = 8,000 – 15,000 EUR. The PoC budget (2,500 EUR) is fully credited.
Monthly Maintenance
Optional, but recommended. AIMOS runs without a maintenance contract too — it is your software.
12-month minimum term
12-month minimum term
6-month minimum term
Flexible Services
For individual requirements outside of the packages.
New skills, connectors, agents, system prompts, ERP integrations
Remote workshops for your team: dashboard, agent management, best practices
Outside business hours, weekends, public holidays
The Real Question
AIMOS is not a chatbot. It is an autonomous system that handles routine work that your employees do manually today — around the clock.
A clerk looks up balances, invoices, payment status in the ERP system
Only during office hours. Vacation/sick leave = downtime.
An employee answers inquiries via email, phone, messenger
Response time depends on availability. Knowledge is lost when staff changes.
AIMOS agent responds instantly, queries ERP, knows the history
No vacation, no sick leave, no knowledge loss. Employees are freed up for more demanding tasks.
AIMOS takes over routine tasks that today keep qualified employees busy below their potential: looking up balances, answering standard inquiries, summarizing documents, following up on appointments. Your employees gain time for what you hired them for — customer relationships, decisions, creative work.
The AIMOS agent has persistent memory: it knows every customer, every preference, every open case. When an employee leaves the company, the knowledge stays — in the agent.
Hardware
AIMOS runs on a single server with an NVIDIA GPU. No cloud, no monthly compute costs. The hardware belongs to you.
One-time purchase — starting at approx. 5,000 EUR
Operating system: Ubuntu 26.04 LTS (recommended) or 24.04 LTS. Power consumption: ~170W average, approx. 45 EUR/month.
Qwen 3.5:27B on RTX 5090 — approx. 40 tokens/sec output, ~500 tokens/sec prompt processing
Single assistant, occasional queries
Multiple departments, regular conversations
Entire company, many conversations, tool chains
A token is the smallest processing unit of the language model — roughly a word or word fragment. A typical Telegram message comprises 30–100 tokens, a detailed agent response 100–300 tokens.
Per agent interaction, the system processes approx. 4,000–6,000 tokens (system prompt + memory + conversation history + response). With 100 interactions per day, that is approx. 500,000 tokens — the GPU is only 10% utilized.
Even with intensive usage with 10 agents, 65% GPU capacity remains free. A SovereignNode is more than sufficiently dimensioned for the entire mid-market. The AI waits for your employees — not the other way around.
No ongoing compute costs: Unlike cloud AI, you do not pay per token. After the one-time hardware purchase (~5,000 EUR), only electricity costs of approx. 45 EUR/month apply — regardless of how many tokens your agents process.
AIMOS is built on a stack of fully open-source components. No proprietary dependencies, no hidden license costs.
| Component | Function | License |
|---|---|---|
| AIMOS | Multi-agent platform | MIT |
| Qwen 3.5 (Alibaba) | Language model (LLM, 27B parameters) | Apache 2.0 |
| SGLang | LLM runtime with RadixAttention (prefix cache) | Apache 2.0 |
| Faster-Whisper | Speech recognition (speech-to-text) | MIT |
| OpenAI Whisper (Model) | Whisper Medium speech model | MIT |
| Piper TTS | Speech synthesis (text-to-speech) | MIT |
| Sentence-Transformers | Embedding model for semantic search | Apache 2.0 |
| PostgreSQL | Relational database (relay + config) | PostgreSQL License (BSD) |
| SQLite | Agent memory (per agent) | Public Domain |
| FastAPI / Uvicorn | Dashboard web server | MIT / BSD |
| Python | Programming language | PSF License (BSD) |
| Ubuntu Linux | Operating system | GPL v2 (free) |
| NVIDIA CUDA Toolkit | GPU driver & compute stack | NVIDIA EULA (free, proprietary) |
| Tailscale | VPN for remote access (optional) | BSD / free for ≤3 users |
The only non-open-source component: NVIDIA CUDA drivers (free but proprietary). All other components are fully open source with permissive licenses. No license fees, no usage restrictions, no vendor lock-ins.
We will advise you free of charge and without obligation on which entry point best suits your company.
Get free consultation