ConfigIQLLM inference sizing
  • Home
  • ESTIMATE
  • Recommend sizing
  • KV cache calculator
  • Predict performance
  • GPU explorer
  • VALIDATION
  • Benchmark comparison
  • COSTINGS
  • Hybrid savings
  • SYSTEM
  • Settings
  • Cost sources
  • Documentation
ConfigIQ v0.4.0 · 009919e

Documentation

ConfigIQ documentation

Guides for using the tools and for running the services behind them. Every page has a stable address you can share.

  • OverviewWhat ConfigIQ is, how the tools fit together, and where to start.
  • In-app helpSearch, ask, and take a tour without leaving the page you are on.
  • User scenariosWorked examples for sizing, predicting, costing, and comparing.
  • REST APIEndpoints, request defaults, response provenance, and examples.
  • MCP serverUse the sizing and pricing APIs from an MCP client.
  • Tested modelsHow measured model and GPU pairs are collected and classified.
  • Pareto validationHow predicted configurations are checked against measured data.
  • Settings and simulatorsSimulator, backend, and version choices, and how shared links behave.
  • CostsHosted, rented, and purchased cost assumptions and comparisons.
  • Model and GPU caveatsKnown model and GPU limits, and when a result is an estimate.
  • ContributingRun the app, build the docs, and prepare a pull request.