Pro · 32 GB
Pro-32: 32 GB high-memory VPS
32 GB NVMe VPS for multi-agent systems and services that hold real state in memory.
Who it's for
The most popular Pro size. 32 GB handles a proper multi-agent crew, a moderate vector database for agent memory, or an app plus its database on one box. Enough headroom that you stop watching the RAM graph.
Typical uses
- ▸Multi-agent crew (5–10 agents) with shared memory
- ▸Vector database (Qdrant/Chroma) for RAG + the app using it
- ▸Web app + PostgreSQL + background workers
- ▸Several bots or pipelines on one server
FAQ
How many agents can Pro-32 run?
Comfortably a crew of 5–10 that call an LLM API, plus a vector store for memory. The limit is usually your API rate limits and budget, not the server.
Can I self-host a small language model here?
Small quantized models on CPU are possible but slow — Pro plans are CPU-only. For real local inference you want a GPU host. Most people call an API and keep the box for orchestration + memory.
Dedicated IP included?
Yes — every Pro plan comes with a dedicated public IPv4 and reverse DNS.
Other Pro sizes