Skip to content

LLM/API Cost Tracker

Cost tracking tool for AI API spend by feature, user and model.

Intermediate16h1–2 people17/20 fit
TL;DR

Build LLM/API Cost Tracker for hackathon builders and small ai teams in ~16h with 1–2 people. Stack: Node.js + LiteLLM + SQLite + React + Charts.

Why it wins: Bill shock is universal and the live attribution demo is easy to grasp.

Build flow

01 · Problem
Teams ship LLM features and only notice the bill when one model, feature or user burned the budget.
02 · Solution
A proxy or SDK wrapper logs every call, attributes tokens and cost by feature, user and model, shows a live dashboard, alerts on budgets and suggests cheaper models or prompt caching.
03 · Build
  1. 1Wrap LLM calls in a proxy
  2. 2Store tokens and cost
  3. 3Build attribution dashboard
  4. 4Add budget alerts
  5. 5Suggest cheaper models
04 · Demo plan
How to demo this to judges:Hit three features with two models → dashboard splits spend → budget alert fires → cheaper-model suggestion shown.

MVP checklist

  • Call logging proxy
  • Spend by feature and user
  • Live cost dashboard
  • Budget alerts
  • Cheaper model tips

Tech stack

TechRole
Node.jsBackend runtime
LiteLLMLanguage model
SQLiteEmbedded database
ReactFrontend UI
ChartsData visualisation

Reference repos

Open-source projects you can build on. Links checked 2026-11-06.

Related ideas

More Devtools
  • Stop runaway agents by capping tokens, cost and steps per task

    Intermediate12h1–2
    Node.jsVercel AI SDKSQLite
    17/20
  • Generate missing unit tests for the code changed in a PR

    Intermediate16h1–2
    LLMNode.jsVitest
    17/20
  • Show a bad code example and get a static analysis rule that finds it everywhere

    Intermediate12h1
    SemgrepLLMNode.js
    17/20