GPT-4.1 token usage increased 48% vs 7-day average
Current impact
$4,200
vs baseline
Forecasted impact / mo
$4,800
if trend continues
Change vs baseline
+48.0%
Input + Output Tokens
Confidence
78%
likely cause attribution
Event Context
Usage Timeline — Input + Output Tokens
Actual vs 7-day rolling baseline
Usage Metric Breakdown
Current vs baseline — all contributing metrics
Input Tokens
2.1M
Baseline: 1.2M
+75%
Output Tokens
760k
Baseline: 410k
+85%
API Requests
6k
Baseline: 3k
+81%
Avg Prompt Length
6k
Baseline: 3k
+75%
Root Cause Analysis
1 confirmed · 2 need confirmation
Requests to /api/reports/summarize: 3,200/day → 5,800/day (+81%)
Avg prompt_tokens per call: 3,200 → 5,600 tokens (+75%)
Interpreted as: Consistent with longer document inputs being passed. Could also reflect a system prompt change — check recent prompt template commits to confirm.
Were prompt templates or document sizes changed around Apr 22?
/api/reports/summarize first appeared in usage data on Apr 22. No deployment context available.
Does Apr 22 correspond to a feature launch or deployment? Annotate to lock this in.
Likely cause
Increased usage of AI Report Summaries feature after product launch
78% confidence · 2 signals awaiting confirmation
Connect a deployment webhook to auto-correlate releases with future spend events.
Suggested Next Steps
Owner: Product Engineering
Review usage spike on /api/reports/summarize endpoint
Check if prompt caching can be enabled to reduce token costs
Confirm with Product Engineering if launch-driven growth is expected
Set a per-endpoint token budget alert at 1.5M tokens/day
Related dimensions