agent calls
19
agent_traces rows
Cost
What the agent spent. Tokens and latency, from agent_traces — there is no
money column, because a price per token is not in the ledger and a figure with no source
is not a figure.
tokens in
442,253
SUM(agent_traces.tokens_in)
tokens out
39,238
SUM(agent_traces.tokens_out)
total latency
192.5s
SUM(agent_traces.latency_ms) · budget 900s per run
By model
| model | calls | tokens in | tokens out | latency |
|---|---|---|---|---|
gpt-oss:120b |
19 | 442,253 | 39,238 | 192.5s |
By run
| run | calls | tokens in | tokens out |
|---|---|---|---|
01M1NCBGSRD0
|
3 | 167,573 | 11,688 |
01M1NC0K91CM
|
3 | 60,227 | 4,922 |
01M1NBX203NC
|
2 | 31,756 | 2,204 |
01M1NBT6RG0V
|
3 | 13,401 | 2,777 |
01M1NBGZ126R
|
2 | 24,579 | 3,106 |
01M1NBAPXKN6
|
3 | 107,235 | 9,336 |
01M1NB5PV636
|
3 | 37,482 | 5,205 |
policy.runComputeBudgetSeconds is 900s. §6.5 records
per-decision cost in decisions.cost_json; this page aggregates the traces.