API
Model Tiers & Routing Architecture
Arcen groups its inference endpoints into three architectural tiers by capability, reasoning density, and cost. Agents dynamically request the exact level of intelligence they need, with predictable routing through the x402 billing layer.
High Model
SOTA & high reasoning
Optimized for complex multi-step software engineering, deep analysis, and fully autonomous agentic workflows.
x402 billing - Higher pre-flight authorization ceiling - the 402 challenge quotes a larger max, sized to the deeper potential token depth.
Balanced Model
Production & complex workflows
The standard for general enterprise tasks, high-accuracy conversation, structured data extraction, and everyday developer tools.
x402 billing - Balanced rate structure with standard per-token USDG settlement per call.
Eco
Low-latency & micro-payments + embeddings
Engineered for high-frequency operations, fast response windows, simple classification, and high-volume vector embeddings.
x402 billing - Micro-priced tokens with no per-call fee - built for high-concurrency loops where cost per token approaches zero.
Infrastructure & monitoring
Monitor gateway liveness, each model tier, and settlement health via public endpoints (no auth). The aggregate returns 200 only when every component is healthy, else 503:
- GET /api/health - aggregate roll-up
- GET /api/health/gateway - x402 gateway & discovery
- GET /api/health/tier-high - High Model tier
- GET /api/health/tier-balanced - Balanced Model tier
- GET /api/health/tier-eco - Eco tier
- GET /api/health/settlement - USDG settlement & ledger
Tier membership reflects the live catalog. Query GET /v1/models for the authoritative list.