running
agent-01
- Framework
- LangChain · CrewAI
- Local model
- Ollama · llama3
- Recommended RAM
- 8 GB
- Storage
- NVMe SSD
Use cases
What you can build
Support agents
An AI agent that receives tickets, searches a knowledge base, and replies to customers automatically.
RAG and knowledge bases
A system that converts your files, PDFs, and websites into a queryable knowledge base.
Code agents
An AI agent that writes, reviews, or debugs code based on natural language instructions.
Task automation
An agent that connects different APIs and performs multi-step tasks automatically.
Features
Why IM Host VPS for AI agents?
Real processing cores
AI agents need fast, genuine CPU cores. Our AMD EPYC and NVMe SSD servers ensure request processing and hook responses with the lowest possible latency.
- Genuine AMD EPYC / Ryzen cores
- NVMe 7x faster than SATA
- ECC DDR4 memory
Enough storage for Vector DBs
Most AI agents need an embeddings database or long-term memory. Our plans provide enough for Chroma, Qdrant, or pgvector.
- 50GB – 2TB NVMe available
- Install any database freely
- Backup every 3 days
Complete data isolation
AI agents often handle sensitive data. KVM VPS means total isolation — no one can access your data or API keys.
- Full KVM virtualisation
- No resource sharing
- Customisable firewall
Run LLMs locally
With AMD EPYC, you can run small to medium LLM models locally using Ollama or LM Studio, eliminating the need for external API calls.
- Ollama, LM Studio support
- llama3, mistral, phi3 models
- No external API fees
Compatible with any framework
LangChain, AutoGen, CrewAI, Flowise or any framework you choose. Full root access means installing any dependency without restrictions.
- LangChain, LangGraph
- AutoGen, CrewAI
- Flowise, n8n, Dify
Stable webhooks and APIs
AI agents frequently receive events from external services. A VPS provides a static IP and permanent webhook URLs that never change.
- Free static IP
- Free SSL with Let's Encrypt
- Reverse proxy with Nginx
What RAM is recommended for an AI agent?
It depends on the framework. LangChain without a local LLM runs on 2GB. With Ollama and a small LLM you need at least 8GB.
Can I run Ollama on a VPS?
Yes. Ollama runs on any Linux VPS. For larger models (7B+) you need substantial RAM. Start with 8GB and scale up as needed.
Does the server support CUDA for GPU?
Our current plans are CPU-only. For GPU inference, we recommend our Intel Line plans with high CPU performance.
How do I secure API keys on the server?
Use environment variables in a .env file outside of git. You can also use Docker secrets or Vault.
Can I run Flowise or Dify?
Yes. Both run on Docker. Install directly with a docker-compose command on any Linux VPS from our plans.
Start running AI agents on your own server
VPS plans from $6/mo. Full root access. NVMe SSD.
