v1.0 is live. Llama 3 & DeepSeek available.

The Open-Source Cloud for AI Agents

Unrestricted Compute. Open Weights. Zero Vendor Lock-In. Stop paying for every comma your AI generates. We provide dedicated infrastructure with predictable, flat-rate pricing.

Deploy Now Read the Docs
agent_config.py
# A drop-in replacement for OpenAI
import openai

client = openai.OpenAI(
    # 1. Point to our open-source cloud
    base_url="https://api.opencloud.diy/v1",
    
    # 2. Use your OpenCloud API key
    api_key="oc_live_...",
)

response = client.chat.completions.create(
    model="meta-llama-3-70b-instruct",
    messages=[
        {"role": "user", "content": "Initialize agent."}
    ]
)

Why Choose Open AI Infrastructure?

The future of artificial intelligence isn't closed behind corporate APIs. It's built on open weights and community-driven innovation.

No "Token Shock"

Traditional AI providers charge by the token, making it impossible to budget for heavy agent workflows or viral traffic spikes. We charge a predictable, flat monthly fee for compute.

True Data Sovereignty

When you use closed APIs, your prompts are often harvested. On our platform, your data stays in your isolated environment. We never read, log, or train on your inputs.

No Censorship

Run the models exactly as their creators intended. No hidden system prompts, no unexpected API deprecations, and no arbitrary rate limits beyond your hardware tier.

Zero-DevOps,
100% FOSS.

We took the headache out of hosting open-weight models. You don't need to configure Kubernetes clusters, manage CUDA drivers, or write Dockerfiles.

  • 1-Click Open Weights

    Instantly spin up dedicated endpoints for the world's most powerful open models including Llama 3, DeepSeek, Mistral & Mixtral, and Qwen.

  • Built on Open Source

    Our entire control plane is built on battle-tested open-source software, ensuring a secure, transparent, and extensible developer portal.

  • Seamless API Integration

    Our API endpoints are a 1:1 drop-in replacement for standard OpenAI/Anthropic SDKs. Just swap the base URL and API key.

$ opencloud models list v1.0.4
meta-llama-3-70b-instruct [READY]
meta-llama-3-8b-instruct [READY]
deepseek-coder-33b-instruct [READY]
mistral-large-latest [READY]
qwen1.5-72b-chat [BOOTING...]
> endpoint: https://api.opencloud.diy/v1/chat

Predictable Monthly Pricing

Choose the compute tier that fits your agent's workload. Pay for the infrastructure, not the intelligence.

Starter

Perfect for hobbyists and single-agent testing.

$30 /mo
  • Access to 8B - 14B models
  • 6,000,000 Compute Credits
  • 50 GB OpenCloud Storage
  • Standard queue priority
Select Starter
Most Popular

Pro

Built for freelancers and multi-agent workflows.

$60 /mo
  • Access to 30B - 70B models
  • 15,000,000 Compute Credits
  • 250 GB OpenCloud + Web Office
  • Priority GPU access
Select Pro

Enterprise

For production apps and heavy agent loops.

$120 /mo
  • Access to 70B+ models
  • 45,000,000 Compute Credits
  • 1 TB OpenCloud Dedicated
  • Dedicated instance routing
Select Enterprise