The OpenSIN Fleet Architecture: 100+ Autonomous Agents
How we built the first open-source AI agent fleet with 100+ specialized agents working 24/7.
← All posts
LLM Architecture
Intelligent Model Routing: Using the Right LLM for Each Task
- Coding tasks → Claude Sonnet 4.6
- Research → Gemini 3.1 Pro
- Creative writing → Gemini 3 Flash
- Simple tasks → Qwen 3.5 or Ollama
By routing simple tasks to cheaper models, we reduce API costs by 60%.
Related Posts
A2A Protocol: The Open Standard for Agent-to-Agent Communication
Deep dive into the Agent-to-Agent protocol that lets AI agents delegate work without human mediation.
Security Architecture: Zero Secrets in Code
How OpenSIN ensures no API keys, tokens, or secrets ever touch the codebase.