eve-eval-trend

1 run · 14 evals · generated 28/07/2026, 14:30

Nothing to compare yet

One run so far, the line starts here.

Judge score 0.76, 11 of 13 evals passed.

What moved

One run so far, so nothing can have moved yet.

11 unchanged · 1 not run · 2 still failing: weather/forecast, weather/units

Per eval

● passed▲ scored✕ failed– skipped· not in that run

Each card covers the last 1 run, oldest → latest.

chat/greeting

● passed
●
29/07/2026, 09:15 · judge 0.94

judge 0.94 · scale 0.60–1.00 · min 0.70

Judged on
“The reply greets the user by name and offers one next step.”
Pass rate
100%
Runs
● 1
Last duration
1.6s

chat/handoff

● passed
●
29/07/2026, 09:15 · judge 0.88

judge 0.88 · scale 0.60–1.00 · min 0.70

Judged on
“The reply names the queue it is handing over to.”
Pass rate
100%
Runs
● 1
Last duration
2.2s

edge/a-deliberately-long-eval-id-that-runs-past-the-card-header-and-keeps-going

● passed
●
29/07/2026, 09:15 · judge 0.93

judge 0.93 · scale 0.60–1.00 · min 0.70

Pass rate
100%
Runs
● 1
Last duration
900ms

intents/handover

● passed
●
29/07/2026, 09:15 · judge 0.95

judge 0.95 · scale 0.60–1.00 · min 0.70

Judged on
“The reply routes to a human when the user asks for one.”
Pass rate
100%
Runs
● 1
Last duration
1.2s

multi/three-conditions

● passed
●
29/07/2026, 09:15 · judge 0.89

judge 0.89 · scale 0.60–1.00 · min 0.70

Judged on
“The reply answers the question that was asked.”
“The reply stays under four sentences.”
“The reply does not invent a source it was not given.”
Pass rate
100%
Runs
● 1
Last duration
4.1s

search/retrieval

● passed
●
29/07/2026, 09:15 · judge 0.72

judge 0.72 · scale 0.60–1.00 · min 0.70

Judged on
“The answer cites at least one retrieved document.”
Pass rate
100%
Runs
● 1
Last duration
3.1s

tools/geocode

● passed
●
29/07/2026, 09:15 · judge 0.91

judge 0.91 · scale 0.60–1.00 · min 0.70

Pass rate
100%
Runs
● 1
Last duration
310ms

weather/forecast

✕ failed
✕
29/07/2026, 09:15 · judge 0.00

judge 0.00 · scale 0.00–1.00 · min 0.70

Judged on
“The reply gives a forecast for the requested day, not the current conditions.”
Pass rate
0%
Runs
✕ 1
Last duration
30.0s

weather/units

▲ scored
▲
29/07/2026, 09:15 · judge 0.58

judge 0.58 · scale 0.50–1.00 · min 0.70

Judged on
“Temperatures are given in Celsius.”
Pass rate
0%
Runs
▲ 1
Last duration
1.9s

edge/no-assertions

● passed
●

no judge assertions on this eval

Pass rate
100%
Runs
● 1
Last duration
120ms

edge/with-image

● passed
●

no judge assertions on this eval

Pass rate
100%
Runs
● 1
Last duration
21.0s

edge/with-output

● passed
●

no judge assertions on this eval

Pass rate
100%
Runs
● 1
Last duration
2.0s

intents/seasonal

– skipped
–

no judge assertions on this eval

Pass rate
—
Runs
– 1
Last duration
—

tools/weather-lookup

● passed
●

no judge assertions on this eval

Pass rate
100%
Runs
● 1
Last duration
240ms

Runs

29/07/2026, 09:15 ● 11 passed▲ 1 scored✕ 1 failed– 1 skipped gpt-5.2-mini · a02c779 · 57.0s

target local · eve 0.27.8 · judge openai/gpt-5.2 · branch main

chat/greeting ● passed↑ Scroll to top

AssertionSeverityScoreMinResult
judge.autoevals.closedQA
“The reply greets the user by name and offers one next step.”		 soft 0.94 0.70 ● pass

chat/handoff ● passed↑ Scroll to top

AssertionSeverityScoreMinResult
judge.autoevals.closedQA
“The reply names the queue it is handing over to.”		 soft 0.88 0.70 ● pass

search/retrieval ● passed↑ Scroll to top

AssertionSeverityScoreMinResult
judge.autoevals.closedQA
“The answer cites at least one retrieved document.”		 soft 0.72 0.70 ● pass

tools/weather-lookup ● passed↑ Scroll to top

AssertionSeverityScoreMinResult
succeeded gate 1.00 — ● pass

tools/geocode ● passed↑ Scroll to top

AssertionSeverityScoreMinResult
judge.autoevals.closedQA soft 0.91 0.70 ● pass

weather/forecast ✕ failed↑ Scroll to top

✕ Timed out after 30000ms waiting for the tool call

AssertionSeverityScoreMinResult
judge.autoevals.closedQA
“The reply gives a forecast for the requested day, not the current conditions.”		 soft 0.00 0.70 ✕ miss

weather/units ▲ scored↑ Scroll to top

AssertionSeverityScoreMinResult
judge.autoevals.closedQA
“Temperatures are given in Celsius.”		 soft 0.58 0.70 ✕ miss

intents/handover ● passed↑ Scroll to top

AssertionSeverityScoreMinResult
judge.autoevals.closedQA
“The reply routes to a human when the user asks for one.”		 soft 0.95 0.70 ● pass

intents/seasonal – skipped↑ Scroll to top

– skipped: only runs between November and January

No assertions recorded.

multi/three-conditions ● passed↑ Scroll to top

AssertionSeverityScoreMinResult
judge.autoevals.closedQA soft 0.89 0.70 ● pass

edge/a-deliberately-long-eval-id-that-runs-past-the-card-header-and-keeps-going ● passed↑ Scroll to top

AssertionSeverityScoreMinResult
judge.autoevals.closedQA soft 0.93 0.70 ● pass

edge/no-assertions ● passed↑ Scroll to top

No assertions recorded.

edge/with-output ● passed↑ Scroll to top

It is currently 14°C and overcast in Bristol, with light rain expected after 4pm. Winds are light from the south-west at around 11 km/h.
AssertionSeverityScoreMinResult
succeeded gate 1.00 — ● pass

edge/with-image ● passed↑ Scroll to top

Here is the fox on a skateboard.
A red fox on a skateboard in a sunlit skatepark, digital illustration.
A red fox on a skateboard in a sunlit skatepark, digital illustration.

image review: most — the fox and the skatepark are there, but its tail is not curled around its paws.

  • image review: most — the fox and the skatepark are there, but its tail is not curled around its paws.
AssertionSeverityScoreMinResult
succeeded gate 1.00 — ● pass