← Back to Tony
Tony AI · Austin Visuals

⚡ API Monitor

Credit balances, token usage estimates, and cost tracking across all Tony's APIs. Updated manually — top up APIs before balance hits red.

✅ All APIs operational. Monitor ElevenLabs credits and OpenAI/Anthropic balances.
API Credit Status
🤖
Anthropic (Claude)  site bill
Primary model — Sonnet 4.6 / Haiku 3.5 · Balance: $77.39
OK
🧠
OpenAI (GPT-5.4)  site bill
Pay-as-you-go · Balance: $46.88
OK
💎
Google Gemini  site bill
Paid tier active · Gemini 2.5 Flash · Pay-as-you-go
OK
🔍
Brave Search  site bill
Web search · API key active · Free tier
OK
🎙️
ElevenLabs (TTS)  site bill
Voice replies · Liam voice · 113K chars remaining of 122K (92%)
OK
📦
Dropbox  site bill
info@austinvisuals.com · Business account · OAuth token active
OK
📧
Gmail (tony@ + matt@)  site bill
Google OAuth · Both accounts connected · Free
OK
⚠️ Credit system note: Tony runs on pre-purchased API credits, not unlimited billing. Balances above are estimates — update manually when you top up. Red = needs attention now. Yellow = top up within a week.
Estimated Token Usage
~85K
Per Day
Active session days
~500K
Per Week
Avg across 7 days
~2M
Per Month
Estimated total
5
Active Cron Jobs
Background tasks running
Token counts are estimates based on current usage patterns (cron jobs every 15 min, email checks, dashboard updates). Heavy research sessions or image generation will spike this significantly.
Approximate Cost (Claude Sonnet + Haiku)
Per conversation day (active) ~$0.50 – $2.00
Background cron jobs only (quiet day) ~$0.10 – $0.30
Per week (mixed active + quiet) ~$3 – $8
Per month (estimated) ~$15 – $35
Heavy research / image gen day ~$5 – $15 spike
💡 Costs are estimates only — actual spend depends on session length, model used (GPT-5.4 costs more than Haiku), and whether image/video generation is used. Tony defaults to cheaper models (Haiku) for background tasks.
Next Actions
🟢 Gemini API Paid tier active — Gemini 2.5 Flash
🟢 ElevenLabs OK — 92% remaining, no action needed
🟢 Anthropic / OpenAI OK — no action needed
🔑 OpenAI Embeddings Quota hit — memory search limited
Last updated Apr 14, 2026 08:34 UTC · Ask Tony to update this page anytime