Skip to main content

Docs

These guides explain Metrum Insights from usage patterns first, then API details. The docs are organized by audience.

Product Documentation

The primary, end-to-end set of customer-facing guides:

  • Quickstart: run your first benchmark on Metrum Insights in under 15 minutes.
  • User Guide: every core workflow, from running benchmarks to KYAI, ASR, VLM, BYOE, server management, and telemetry.
  • Feature Reference: per-feature configuration lookup across vLLM, SGLang, TensorRT-LLM, KYAI, Sizer, Benchmarks, RBAC, cloud provisioning, and more.
  • Admin & Configuration Guide: identity, RBAC, billing, backups, GDPR, and cloud provisioning for platform admins.
  • API Reference: REST and RPC surface, authentication, pagination, and v3 -> v4 column mapping.
  • Hardware Compatibility: supported GPU SKUs, TPUs, and accelerators across workload types with driver and firmware requirements.
  • Sizer Usage Guide: pick the right hardware for an LLM workload before you deploy.
  • Release Notes - v4.0: what's new, what's changed, and what's been deferred.

User Docs

For Web UI users and benchmark operators who need deeper engineering context.

  • Benchmark Lifecycle: end-to-end process from benchmark question to packaged results.
  • Hardware Sizer (engineering reference): the engineering view of how Metrum's benchmark dataset, model signals, workload targets, and owned inventory turn into GPU deployment recommendations.
  • Workflow Examples: starter workflow and comparison guides for performance, quality, modality, and engine studies.
  • Benchmark Planning: model, precision, modality, ISL/OSL/concurrency sweeps, tools, and acceptance criteria.

Admin Docs

For operators installing and maintaining Metrum Insights.

Developer Docs

For engineers integrating with or extending Metrum Insights.

API Reference

The auto-generated, runbook-style API references: