Usage sub-tab

The third sub-tab of an AI agent. Shows this agent’s token consumption and cost trends.

When to use: “How much did this cost this month”, “Why did it suddenly burn so much”, “Should I upgrade my plan”.

Where to find

OC-1 (or OC-2 / HM-3) tab → Usage sub-tab

Token usage trend for the last N days (bar or line chart). X-axis is date, Y-axis is token count.

How to read:

Steady trend = normal use
Sudden spike = AI might be in a loop, scheduled task issue, or you’re doing heavy chats
All zero = AI not used (normal)

Token counts split by use:

Category	Example
Conversation	Direct user-AI interaction
Scheduled Tasks	AI’s autonomous schedules
Tool Use	Internal back-and-forth when AI uses tools

Cumulative USD cost based on current model rates (prompt + completion).

OC / HM let you pick different models, and cost differs hugely:

Model	input/M tokens	output/M tokens	Best for
claude-opus-4-7	High	High	Complex reasoning, long tasks
claude-sonnet-4-6	Medium	Medium	General chat, balanced
kimi-k2.6	Low	Low	High volume, simple tasks
MiniMax-M2.7	Low	Low	Chinese chat, low latency

Actual rates per provider’s current pricing. Switching model → token count doesn’t change, but USD cost translation does.

To switch model: go to Settings sub-tab and adjust the model dropdown.

Troubleshoot in this order:

Check Activity sub-tab for recently failed scheduled tasks (AI might be retrying)
Check recent days’ chats: any unusually long prompts or many interactions
Check model selection: did you accidentally switch to an expensive model
Pause + observe: if uncertain, pause for 24 hours and watch the trend chart flatten

Shape	Usually means
Flat baseline + occasional spikes	Normal — spikes are when you do big things
Steady climb	Schedules adding up daily, or chats getting longer
Sudden cliff	AI was paused or restarted
Permanently elevated plateau	Possibly a loop; pause agent to investigate