evals
Run evals on Jetty
v1 · May 2026 · jettyio · auto-evolved runbook (weakest-axis targeting)

hermes · GLM 5.1

Zhipu's GLM 5.1. The steadiest v1 climber: 24, 30, 35, 36.5 with no grade inflation.

agent hermes · model openrouter/z-ai/glm-5.1

Judge, best round
27.3/40
round v2
Self-score, best round
36.5/40
inflation +9.2
Rounds
10
Jetty trajectories
hermes · GLM 5.1 best round
best round v2 · 4b324d80 ↗

Self-score vs judge, by axis

Best round v2, 0–10 per axis

agent self-scoreexternal judge

Judge: The bicycle is solid, with spoked wheels, a red frame, handlebars, a seat and chain. The pelican sits on the seat with its orange legs reaching the pedals and its wing extended toward the handlebars. The weakest part is the pelican's head: the throat pouch is a huge orange blob hanging off the neck rather than under the beak, which makes the anatomy odd. The wing rendering is also somewhat messy, with stray overlapping feather shapes near the handlebars.

Score by round

Out of 40. Each round starts from the previous round's best SVG.

self-scorejudge

Every round

Click a drawing to open it in the head-to-head viewer. The outlined card is the agent's best round.

hermes · GLM 5.1 round v1
v1judge 28.7 · self 32 c1f2da04 ↗runbook
hermes · GLM 5.1 round v2
v2 · bestjudge 27.3 · self 36.5 4b324d80 ↗runbook
hermes · GLM 5.1 round v3
v3judge 28.8 · self 36 f443b217 ↗runbook
hermes · GLM 5.1 round v4
v4judge 28.7 · self 33 7ce1468b ↗runbook
hermes · GLM 5.1 round v5
v5judge 27.7 · self 33 ab8cb01d ↗runbook
hermes · GLM 5.1 round v6
v6judge 25.7 · self 34.5 9fd19804 ↗runbook
hermes · GLM 5.1 round v7
v7judge 25 · self 36 83e8efe2 ↗runbook
hermes · GLM 5.1 round v8
v8judge 24.7 · self 34 21b9b8c1 ↗runbook
hermes · GLM 5.1 round v9
v9judge 24.2 · self 33 9f9e50be ↗runbook
hermes · GLM 5.1 round v10
v10judge 25.7 · self 36.5 fff60669 ↗runbook

Scores per round

RoundPelicanBicycleCompositionPolishSelf totalJudge totalJudge spreadMinutesTrajectory
v17self 87.2self 87.5self 87self 83228.70.5–c1f2da04 ↗
v26.5self 9.57.3self 9.57self 8.56.5self 936.527.30.5–4b324d80 ↗
v37self 97.3self 87.5self 107self 93628.80.5–f443b217 ↗
v47self 87.2self 97.5self 87self 83328.73.5–7ce1468b ↗
v56.3self 97.3self 87.2self 86.8self 83327.70.5–ab8cb01d ↗
v65.8self 8.57self 9.56.7self 8.56.2self 834.525.73.5–9fd19804 ↗
v75.5self 97self 96.5self 96self 936250–83e8efe2 ↗
v85.5self 97self 86self 96.2self 83424.70.5–21b9b8c1 ↗
v95.5self 87self 95.7self 86self 83324.20.5–9f9e50be ↗
v106self 9.57.2self 96.2self 96.3self 936.525.71.5–fff60669 ↗

Want this kind of eval for your own agents?

Every run on this site is a Jetty runbook: versioned instructions, a pinned sandbox, and a trajectory you can inspect. Point Jetty at your task and get the same receipts.