© 2026 Woonggi Min ·notes / source

AI infrastructure

The Load Balancer

Three projects connecting models, accounts, and inference requests.

  • TokenhubPrivate

    An AI gateway connecting accounts across model providers and distributing requests.

    Connection & setup
    Connect
    Connect existing model clients through OpenAI- and Anthropic-compatible APIs.
    Configure
    Manage account connections and client keys, and inspect routing status for each model.
    Token Hub wordmark, sidebar, and account availabilityToken Hub wordmark, sidebar, and account availabilityToken Hub model providers and availabilityToken Hub model providers and availabilityToken Hub request search and usage statisticsToken Hub request search and usage statistics

  • kiro-lbminpeter

    Multi-account routing built on kiro-gateway, with OpenAI- and Anthropic-compatible APIs.

    Connection & setup
    Connect
    Run a standalone binary or Docker container, then point a compatible client at its API.
    Configure
    Manage account sign-in, API keys, load balancing, and concurrency from the dashboard.
    KiroLB wordmark, sidebar, and request volume chartKiroLB wordmark, sidebar, and request volume chartKiroLB account status and remaining quotaKiroLB account status and remaining quotaKiroLB concurrent request and queue timeout settingsKiroLB concurrent request and queue timeout settings

  • InferXinferxhq

    An inference exchange connecting spare provider quota through a shared API.

    Connection & setup
    Connect
    Suppliers connect supported accounts; consumers call models with an API key.
    Configure
    Manage provider connections and API keys, and track usage, credits, and supplier earnings.
    InferX wordmark, sidebar, and workspace overviewInferX wordmark, sidebar, and workspace overviewInferX credit balance and usage ledgerInferX credit balance and usage ledgerInferX requests, token usage, and earned credits by modelInferX requests, token usage, and earned credits by model