Live demo · sample data
Acme AI — all providers
This is the real dashboard with made-up numbers. Try the pause button on the spike.
Spend 7.3× normal on support-bot (prod)
$17.27 in the last hour on Anthropic. Normal for this hour is about $2.36. A loop, a leaked key or a traffic spike?
Spent in November
$987
Day 11 of 30
Forecast to month end
$2,888
$612 under budget
Per day lately
$96.53
Last 7 full days
Last hour
$20.98
1 spike found
Spend this month
Actual Forecast Budget
Budgets
- All AI spend$987 of $3,500 · forecast $2,888
- Anthropic$590 of $2,200 · forecast $1,719
- Vercel$127 of $450 · forecast $381
Providers
- AnthropicUpdated 1 min ago$590
- OpenAIUpdated 2 min ago$242
- VercelUpdated 3 min ago$127
- OpenRouterUpdated 4 min ago$28
Top keys and projects
- support-bot (prod)$328
- coding-agent$206
- Search · api$172
- dashboard-app$89
- Mobile app · app$70
- batch-tagging$56
- marketing-site$39
- research-scripts$28
Last 24 hours
per hour, estimatedRate-limit headroom
- Claude Sonnet · input tokens/min14% leftpeak 1.7M / 2.0M tokens/min
- Account credits used21% leftpeak 118 / 150 USD
- Claude Opus · input tokens/min59% leftpeak 410k / 1.0M tokens/min
- Claude Sonnet · requests/min64% leftpeak 1.4k / 4.0k requests/min
Model-swap savings
Estimate from your own token mix this monthClaude Sonnet → Claude Haiku
−$67
$101 → $34 for 23M tokens
GPT-5 → GPT-5 mini
−$42
$52 → $10 for 24M tokens
Claude Opus → Claude Sonnet
−$20
$50 → $30 for 7.0M tokens
Only worth it where the cheaper model is good enough. Test on a slice of traffic first.
Recent alerts
- Spend 7.3× normal on support-bot (prod)$17.27 in the last hour. Sent to email, Telegram and phone.12 min ago
- Only 14% left: Claude Sonnet · input tokens/minRequests may start failing with rate-limit errors at peak.2 h ago
- Anthropic: forecast $1,719 of $2,200$590 spent so far (27%). In the daily digest.6 h ago
- Daily digest sentYesterday's spend, month to date and forecast, at 9:00 your time.6 h ago