Sanctum Router
AI model routing service. Unified API for DeepSeek, OpenRouter, and OpenCode models with intelligent routing, rate-limiting, and load balancing.
Multi-Provider Routing
Intelligent request routing across DeepSeek, OpenRouter, OpenCode, and custom model endpoints with failover, latency-based selection, and cost optimisation.
- Unified OpenAI-compatible API across all providers
- Latency-based and cost-aware routing strategies
- Automatic failover on provider degradation or outage
- Provider weighting for gradual traffic migration
Sovereignty Angle
You choose which providers to trust and in what order. No single vendor lock-in — if a provider becomes unsuitable, reroute with a config change, not a migration.
Model Catalogue
A browsable, searchable catalogue of 50+ AI models with capability descriptions, pricing metadata, and performance benchmarks — all in one place.
- 50+ indexed models with metadata and capability tags
- Search and filter by provider, capability, and cost tier
- Live pricing and rate-limit information per model
- Performance benchmarks and latency estimates
Sovereignty Angle
The model catalogue is a first-party data file — no external API is queried to display model information. You always know what you're calling and what it costs.
Rate Limiting
Granular rate limiting and load balancing at the user, API key, and provider level — protecting both your upstream provider limits and your own infrastructure.
- Per-user, per-key, and per-provider rate limit tiers
- Token-based and request-based throttling
- Distributed rate limiting via Redis backend
- Load balancing across multiple provider accounts
Sovereignty Angle
Rate limits are enforced by your own router on your own infrastructure. No external gateway vendor sits between your users and the models.
MCP Integration
Expose the full router as an MCP (Model Context Protocol) server — AI agents can query available models, check routing status, and manage configuration through tool calls.
- MCP server exposing router management as tools
- Agent-driven model selection and routing configuration
- Real-time provider health and latency queries
- Seamless integration with Sanctum Chat and external MCP clients
Sovereignty Angle
Your AI agents manage their own routing via an open protocol you control. No proprietary agent-to-router middleware — just standard MCP on your infrastructure.