OpenClaw
Open-source personal AI assistant that connects to WhatsApp, Telegram, Discord, Slack, and more. Runs 24/7. Choose cloud AI or run locally with Ollama.
Use cases
Deployment options
OpenClaw can run with cloud AI or with Ollama on your own equipment. Choose based on cost, data sensitivity, and control.
Cloud AI
Connect to Anthropic, OpenAI, or other cloud APIs
Pricing: OpenClaw is free (MIT). You pay for the cloud API: Anthropic, OpenAI, etc. Per-token pricing applies. Use our Cost Calculator to estimate.
Pros:
- Best model quality (Claude, GPT-4)
- No hardware to manage
- Fast to get started
- Same agent features: memory, skills, multi-channel
Cons:
- API costs scale with usage
- Data sent to cloud provider
- Ongoing per-token bills
Data handling: Data flows to your chosen cloud provider. Check provider terms (Anthropic, OpenAI) for training and residency.
Ollama (on your equipment)
Run models locally with Ollama. No API bills.
Pricing: OpenClaw is free. Ollama is free. Cost is hardware (GPU) and power. See our Self-hosted GPU Comparisons for £/1M token estimates.
Pros:
- No per-token costs
- Data never leaves your machine
- Offline capable
- Full control
- Same 5,700+ ClawHub skills
Cons:
- Upfront hardware cost
- You manage updates and security
- Model quality depends on GPU
- Requires 5GB+ VRAM (8B) to 40GB+ (70B)
Data handling: All data stays on your hardware. Complete data sovereignty.
Business fit
Pros
- Open source
- Self-hostable or cloud
- One assistant for many channels
- Extensible
- 24/7 availability
Cons
- Requires setup (Node 22+, config)
- Cloud mode: API costs
- Self-hosted: hardware + maintenance