s

sie

Developed by superlinked

SIE (Superlinked Inference Engine) is an open-source, self-hosted inference engine designed for AI agent tasks. It serves all models (search, doc-to-markdown, structured output, content safety, and agent loops) from a single cloud cluster via an OpenAI-compatible API. It supports on-demand loading of 100+ models with LRU eviction and ships with a production stack including load balancing, KEDA autoscaling, and Terraform modules.

  • Unified OpenAI-compatible API
  • On-demand loading & LRU eviction for 100+ models
  • Production-grade KEDA autoscaling & Terraform modules
  • Pre-configured tasks (Search, OCR, Structured Output, Guardrails)
  • Seamless integration with major LLM and Vector DB frameworks
web