Agent Cost & Tokens
Per-discipline token spend and USD cost across the build crew — cheap models (haiku, swarmiq-local) carry the bulk volume while opus is reserved for review-grade work. ● live from agent_runs
Spend totals
Aggregate tokens and metered USD across every agent run.
Spend over time
Aggregate USD per run-window, premium (opus) vs cheap (haiku / swarmiq-local). The swarm leans hard on cheap/free inference.
Per-agent cost
Model, token throughput, USD, and the cost trend per discipline. Free local inference (swarmiq-local) reads $0.
Tokens in
1.84M
Tokens out
0.31M
Cost trend
Tokens in
3.64M
Tokens out
0.81M
Cost trend
Tokens in
2.41M
Tokens out
0.54M
Cost trend
Tokens in
1.56M
Tokens out
0.30M
Cost trend
Tokens in
1.41M
Tokens out
0.28M
Cost trend
Tokens in
1.12M
Tokens out
0.27M
Cost trend
Tokens in
0.98M
Tokens out
0.18M
Cost trend
Tokens in
1.18M
Tokens out
0.23M
Cost trend
Tokens in
1.29M
Tokens out
0.30M
Cost trend
Tokens in
0.64M
Tokens out
0.14M
Cost trend