InfraX Router

Route smarter. Ship faster.

Infrastructure routing for LLMs. No password — your API key is your account.

Your API key

Copy it now. For security we don’t show the full key again.

Already have a key? → Dashboard
API docs
Blog

InfraX Router vs OpenRouter Auto-Route

How our task-aware routing compares to naive single-model routing

Based on real benchmarks
You save 92.3%
per 1M tokens $0.60
InfraX Router · $0.05/1M task-aware routing across 6 models
Naive baseline · $0.65/1M always the priciest model (llama-3.1-70b)

live · /billing/savings/public

InfraX Router cost savings vs naive baseline Two side-by-side bars. The naive baseline bar (red) is tall at $0.65 per 1M tokens. The InfraX Router bar (green) is tiny at $0.05 per 1M tokens — a 92.3% saving, shown as a strike-through badge between them. $0.65 /1M NAIVE always priciest model llama-3.1-70b $0.05 /1M INFRAX smart routing 6 models in pool 92.3% SAVED $0.60 / 1M

Methodology: Naive baseline = always the most-expensive model enabled in the pool. InfraX Router = task-aware routing picks the cheapest capable model per task type (short_chat 35%, coding 20%, general 20%, long_reasoning 15%, math 10%). Numbers are fetched live from GET /billing/savings/public — refresh the page to re-query. This is a theoretical maximum, not a guarantee; real savings depend on your task mix, cache hits, and quality thresholds.