Agent roles & token allocation
Allocation % should sum to 100 (auto-normalized on compute).
Token flow diagram (flat example)
Flat topology: orchestrator + N workers (allocation weighted).
Topology comparison (flat · hierarchical · pipeline)
| Topology | Total tokens/task | Max parallel tasks | Cost/task ($) | Monthly tasks (budget) |
Tasks/month vs agent count (flat topology)
Recommendation
Based on budget and overhead, pipeline topology yields lowest cost per task for your configuration.
FAQ
- How is inter-agent overhead calculated? Overhead = avg messages per task × avg tokens per message. Added to total tokens per task.
- Which topology uses least tokens? Pipeline typically uses fewest tokens because each agent only sees previous output, reducing context.
- Can I set a monthly dollar budget? Yes. Select a model and enter monthly $ budget; the tool converts to token budget using model price.