Tokens & Cost
Invoked tracks token usage and dollar cost for every run, broken down by model and provider, so you can see exactly what your agents spend and where.
Every agent run consumes tokens — the units LLM providers bill by. Invoked tracks token usage and translates it into dollar cost for each run, so spend is never a mystery.
What gets tracked
For each run, Invoked records:
- Input and output tokens consumed by the model.
- Dollar cost, computed from the model's published pricing.
- The model and provider used, so multi-model workspaces stay attributable.
This appears inline in Run History and rolls up into the Dashboard, where a token-usage card shows spend over time.
Why it matters
Agentic workflows can quietly get expensive — a multi-step run with tool calls and retries burns far more than a single prompt. Surfacing cost per run turns that from a surprise on your provider invoice into something you can see, compare, and optimize as you build.
Cost is computed locally from token counts and model pricing. Your prompts and run contents stay on your machine.
For teams
On the Team plan, token and cost data rolls up to workspace-scoped stats, giving admins a spend and ROI view across the team without exposing individual run contents.