Run started 2026-10-18 09:09 UTC
· completed 09:23 UTC
Headline Metrics
A case passes when (Correctness or String in Response) and Relevance and Tools pass.
Scorers that had nothing to judge are ignored; cases where no scorer produced a result are
counted as no data and excluded from every pass rate.
First Token Time
Time to the first response token (TTFT) — how long before the user sees anything.
mean
94.19s
p50
88.20s
p90
145.06s
p99
177.47s
min
51.58s
max
181.07s
std
39.16s
First Insight Time
Time to the agent's first update, such as a tool call or a reasoning step.
mean
8.16s
p50
6.75s
p90
12.80s
p99
16.35s
min
2.96s
max
16.74s
std
4.22s
Case Duration
Wall-clock time the eval harness measured for the whole case, including scoring and gRPC overhead. Always at least the agent's own time_total.
mean
108.67s
p50
95.84s
p90
153.98s
p99
210.75s
min
57.50s
max
218.35s
std
43.55s
LLM Time
Total time inside LLM spans, summed across every LLM call in the case.
mean
68.19s
p50
54.68s
p90
115.10s
p99
147.60s
min
40.42s
max
151.21s
std
34.55s
Tool Time
Total time executing tools — database queries, API calls and the like.
mean
34.95s
p50
30.94s
p90
52.91s
p99
60.75s
min
11.36s
max
61.62s
std
16.03s
Other Time
Agent time outside the LLM and tools: orchestration overhead.
mean
0.68s
p50
0.00s
p90
3.39s
p99
3.98s
min
0.00s
max
4.04s
std
1.51s
LLM Time Share
LLM time as a percentage of the agent's total span duration. This is the spread of per-case percentages, not the run-level ratio.
mean
68.9%
p50
65.2%
p90
80.8%
p99
86.5%
min
53.9%
max
87.1%
std
10.6%
Tool Time Share
Tool time as a percentage of the agent's total span duration. This is the spread of per-case percentages, not the run-level ratio.
mean
36.4%
p50
40.2%
p90
47.6%
p99
57.2%
min
16.4%
max
58.3%
std
12.9%
Tool Calls
Number of tool spans the agent produced for the case.
mean
12.18
p50
11.00
p90
18.00
p99
19.80
min
7.00
max
20.00
std
4.24
sum
134.00
Total Tokens
LLM tokens consumed across every call in the case (prompt plus completion).