Where these numbers come from. Grok Bot does not tell the box what a run costs or how many tokens it used, so there is no real bill here. Counted: runs (from log-run.sh), run minutes (only when log-run.sh gets the optional 5th argument, seconds), and actions (bot activity, kept 7 days). The estimate column multiplies those by your rates in Settings: $0.02/run, $0.01/run minute, $0.001/action.
| Bot / routine | Runs ▼ | Failed | Run minutes | Actions | Estimate | Model calls | Tokens in / out | Token cost (est.) | |
|---|---|---|---|---|---|---|---|---|---|
| ▾ Inbox Bot | 338 | 25 | 969.4 | 159 | ~$16.61 | 1255 | 30.3M / 1.3M | ~$7.14 | |
| Inbox triage rt_demo_triage | 336 | 24 | 963.7 | - | ~$16.36 | 1249 | 30.2M / 1.3M | ~$7.11 | |
| Waiting-on digest rt_demo_digest | 2 | 1 | 5.7 | - | ~$0.0972 | 10 | 194k / 9k | ~$0.0462 | |
| ▾ Deploy Bot | 56 | 4 | 156.4 | 148 | ~$2.83 | 196 | 4.6M / 201k | ~$1.09 | |
| CI watch rt_demo_ci | 56 | 4 | 156.4 | - | ~$2.68 | 196 | 4.6M / 201k | ~$1.09 | |
| ▾ Research Bot | 9 | 1 | 53.3 | 152 | ~$0.8648 | 46 | 1.2M / 46k | ~$1.85 | |
| Morning brief rt_demo_brief | 7 | 1 | 26.6 | - | ~$0.4065 | 26 | 720k / 23k | ~$0.1622 | |
| Friday deep dive rt_demo_deep | 2 | 26.6 | - | ~$0.3063 | 20 | 447k / 23k | ~$1.69 | ||
| ▾ Garden Bot | 7 | 2 | 19.5 | 141 | ~$0.4758 | 23 | 714k / 20k | ~$0.1588 | |
| Watering check rt_demo_water | 7 | 2 | 19.5 | - | ~$0.3348 | 23 | 714k / 20k | ~$0.1588 | |
| Total estimate, last 30 days | ~$20.79 | ||||||||
Estimate only. Check your real bill with your provider.
Tokens are estimated from transcripts. The box counts transcript bytes (about 3.9 bytes per token, measured against the o200k tokenizer) for each model call: input = system prompt + everything earlier in the conversation, output = the reply. It ignores prompt caching and hidden tool context, so treat it as a rough size, not a bill. Per routine = that bot's tokens during the run (its logged duration, else the 10 minutes before it finished).