← campaign 697e79b75e41, execution 1

Comparison: campaign 697e79b75e41, execution 1 against campaign 13fda3099afa, execution 1

Newer: campaign 697e79b75e41, execution 1, on gpt-5.6-terra. Earlier: campaign 13fda3099afa, execution 1, on gpt-5.6-terra. Compared in the order the maintainer chose. Across campaigns only deltas against each execution's own baseline are compared.

Pins that differ

Each arm against its own baseline

CategoryScenarioArmModelMetriccampaign 13fda3099afa, execution 1campaign 697e79b75e41, execution 1Change
CS1@1.0baseline-docsgpt-5.6-terraM-Q1−0.4 pp n = 3 / 3, range 99.4%–100.0% / 100.0%–100.0% within variance+1.0 pp n = 3 / 3, range 99.4%–99.4% / 96.4%–99.4% within variance+1.4 pp
CS1@1.0baseline-docsgpt-5.6-terraM-K1+0.0440 EUR n = 3 / 3, range 0.6621 EUR–0.7858 EUR / 0.6352 EUR–0.7702 EUR within variance+0.0959 EUR n = 3 / 3, range 0.7888 EUR–0.8635 EUR / 0.6958 EUR–0.7504 EUR beyond variance+0.0519 EUR
CS1@1.0wingfoilgpt-5.6-terraM-Q1−0.2 pp n = 3 / 3, range 99.4%–100.0% / 100.0%–100.0% within variance+1.4 pp n = 3 / 3, range 99.4%–100.0% / 96.4%–99.4% within variance+1.6 pp
CS1@1.0wingfoilgpt-5.6-terraM-K1+0.6237 EUR n = 3 / 3, range 1.1666 EUR–1.4108 EUR / 0.6352 EUR–0.7702 EUR beyond variance+0.4619 EUR n = 3 / 3, range 1.0827 EUR–1.2612 EUR / 0.6958 EUR–0.7504 EUR beyond variance−0.1618 EUR
DS2@1.0baseline-docsgpt-5.6-terraM-Q1±0.0 pp n = 1 / 1 preliminary±0.0 pp n = 1 / 1 preliminary±0.0 pp
DS2@1.0baseline-docsgpt-5.6-terraM-D3±0 n = 1 / 1 preliminary±0 n = 1 / 1 preliminary±0
DS2@1.0wingfoilgpt-5.6-terraM-Q1±0.0 pp n = 1 / 1 preliminary±0.0 pp n = 1 / 1 preliminary±0.0 pp
DS2@1.0wingfoilgpt-5.6-terraM-D3±0 n = 1 / 1 preliminary±0 n = 1 / 1 preliminary±0
ES8@1.0baseline-docsgpt-5.6-terraM-E1−10 n = 1 / 1 preliminary−10 n = 1 / 1 preliminary±0
ES8@1.0wingfoilgpt-5.6-terraM-E1−10 n = 1 / 1 preliminary−10 n = 1 / 1 preliminary±0
FS3@1.0baseline-docsgpt-5.6-terraM-F1+20.0 pp n = 1 / 1 preliminary+20.0 pp n = 1 / 1 preliminary±0.0 pp
FS3@1.0wingfoilgpt-5.6-terraM-F1+40.0 pp n = 1 / 1 preliminary±0.0 pp n = 1 / 1 preliminary−40.0 pp

Not compared

On one side only

None.