CarbonHelm
LLM Cost Attribution
See WHERE your AI spend goes. No SDK required.
API Keys
Or: Paste Billing Export
Or: Manual Entry
Keys are used locally in your browser and never transmitted to any server.
Total Monthly AI Spend
$2,047
+12.3% vs last month
Spend by Provider
Spend Trend — Last 30 Days
Cost per 1K Tokens by Model
Most Expensive Model
GPT-4o
$1,124/mo — 55% of total spend — 2.4M tokens processed
Total Potential Savings
$528
per month — 25.8% of current spend
Implementation Guide
Step 1: Switch /api/classify to Claude Haiku (15 min, no code change needed if using LiteLLM)
Step 2: Enable prompt caching on /api/chat system prompts (add cache_control header)
Step 3: Batch /api/summarize calls with a 60-second window
Step 4: Set up budget alerts at $60/day threshold
Cost Per User Breakdown
847
Active Users
$1.95
Avg Cost/User
12
Flagged Users
Top Users by AI Cost
Recommendations
Rate Limiting: Cap power users at 200 AI calls/day to reduce spend by ~$180/mo
Tier-Based Access: Reserve GPT-4o for Pro plan users; route Free users to Haiku. Saves ~$340/mo
Caching: 23% of user queries are repeated within 1 hour. Cache responses to save ~$95/mo
Set Budget Alert
Active Alerts
Alert History
Choose Your Plan
Start free. Upgrade when your AI spend demands it.
Free
$0
  • Analyze up to $500/mo in AI spend
  • 3 provider connections
  • Basic cost dashboard
  • 7-day data retention
  • Community support
Enterprise
Custom
  • Everything in Pro
  • Team access & RBAC
  • SSO / SAML integration
  • REST API access
  • Unlimited data retention
  • Dedicated support
  • Custom integrations