ConfigIQ

LLM inference sizing, GPU comparison, and cost modeling for engineers and infrastructure teams.

Live

• configiq.xyz (latest release)
• configiq.dev (latest commit)

Built with Next.js + PatternFly, powered by our AISimulators REST API.

What it does

Tool	Description
Performance	Fast GPU memory and cost estimate from model + load profile
Recommend Sizing	Detailed sizing with batching, quantization, and cost modeling
KV Cache Calculator	Memory breakdown and KV cache capacity analysis
GPU Explorer	Compare GPUs across memory, throughput, cost, and availability
Hybrid Savings	Model cost savings across cloud, on-premise, and hybrid strategies
Routing Economics	Analyze request routing between model tiers
Cluster cost	Estimate costs for multi-node GPU clusters

Getting started

Prerequisites

• Node.js >= 20.19.0
• npm >= 10

Setup

git clone https://github.com/redhat-performance/configiq.git
cd configiq
npm install
cp .env.example .env.local
npm run dev

App runs at http://localhost:3000.

Available commands

npm run dev          # Start dev server (http://localhost:3000)
npm run build        # Production build
npm run type-check   # TypeScript check without building
npm run lint         # ESLint
npm test             # Vitest test suite

Tech stack

Layer	Technology
Framework	Next.js 16 App Router + TypeScript
UI	PatternFly v6
Backend APIs	AISimulators for GPU sizing and estimation; aicostings for pricing

Project structure

app/                  Next.js App Router pages
  layout.tsx          Root layout, fonts, PatternFly CSS imports
  page.tsx            Homepage
  recommend/          Recommend sizing tool
  kv-cache/           KV Cache Calculator
  predict/             Predict performance
  gpu-explorer/       GPU Explorer
  hybrid-savings/     Hybrid Savings
  routing/            Routing Economics
  settings/           App settings
  api/                Next.js same-origin proxies and application APIs
    recommend/        POST — GPU sizing via AISimulators /recommend
    predict/          POST — GPU performance via AISimulators /predict
    memory/           POST — memory breakdown via AISimulators /memory
    gpus/             GET — GPU catalog via AISimulators /systems
    catalog/          GET — combined systems, models, and backends catalog
    estimate/         POST — compatibility alias for predict
    hf-config/        GET — Hugging Face model config lookup
    health/           GET — health check
    costings/         Pricing-service proxies
    metrics/          Application metrics
components/
  layout/
    AppShell.tsx      Top-nav masthead + sidebar navigation
lib/
  api/                AISimulators and aicostings API clients
docs/                 Architecture docs and ADRs

Contributing

Before opening a PR

Run these checks before opening a PR:

npm run type-check   # Must be clean
npm run lint         # Must be clean
npm run build        # Must succeed

Code conventions

1. GPU math belongs in aisimulators — never write sizing formulas inside React components.
2. Pricing belongs in the aicostings service — never add costing inside React components.
3. PatternFly only — do not add Tailwind, shadcn/ui, or any other component library.
4. Sentence case everywhere — no title case in headings or labels.
5. Server components by default — add "use client" only when needed.
6. No any types — TypeScript strict mode is enforced.