Cost Tracking

Track token usage and costs across all agents. Optional and off by default.

Setup

cost.py

from veska import CostTracker

tracker = CostTracker(enabled=True)

# Or enable later
tracker = CostTracker()
tracker.enable()

Recording usage

Usage is recorded automatically when using the Orchestrator with tracking enabled. You can also record manually:

record.py

tracker.record(
    agent_name="researcher",
    model="claude-sonnet-4-6",
    input_tokens=1500,
    output_tokens=800,
    task_id="task_001",
)

Querying costs

query.py

# Total tokens
print(tracker.total_tokens)
# {"input": 15000, "output": 8000, "total": 23000}

# Total cost
print(f"Total: ${tracker.total_cost:.4f}")

# Per-agent breakdown
for agent, cost in tracker.get_agent_costs().items():
    print(f"  {agent}: ${cost:.4f}")

Custom pricing

pricing.py

# Override default pricing (per 1M tokens)
tracker.set_pricing("claude-sonnet-4-6", input_price=3.0, output_price=15.0)
tracker.set_pricing("gpt-4o", input_price=2.50, output_price=10.0)

With Orchestrator

orch.py

orchestrator = Orchestrator(
    model="claude-sonnet-4-6",
    agents=[researcher, writer],
    tracking={"enabled": True},
)

result = orchestrator.run("Write a report")
# Cost tracking happens automatically

Persistence

persist.py

# Save cost data to your own storage
def save_costs(record):
    db.insert("costs", record)

tracker.set_storage(save_costs)