32 trials · total cost $80.59 · not on this page: 11 attempts that ended on provider rate limits or credit ($1.72, re-run in exp_cover_interlock_openai) and the duplicate run of Sol T3 trial 2 ($0.50); both are in the dataset's episodes table · click a trial to step through it
| agent | task | trials | harness errors | submitted | all pass | mean end score | cost | cost / trial |
|---|---|---|---|---|---|---|---|---|
| argon | t3_light | 3 | — | 3 | 0 / 3 | 0.776 | $7.6748 | $2.5583 |
| argon | t4_takeover | 3 | — | 3 | 0 / 3 | 0.981 | $6.0905 | $2.0302 |
| argon | all tasks | 6 | — | 6 | 0 / 6 | 0.879 | $13.7653 | $2.2942 |
| astra | t3_light | 3 | — | 3 | 1 / 3 | 0.866 | $10.9353 | $3.6451 |
| astra | t4_takeover | 3 | — | 3 | 2 / 3 | 0.99 | $8.0980 | $2.6993 |
| astra | all tasks | 6 | — | 6 | 3 / 6 | 0.928 | $19.0332 | $3.1722 |
| fable | t3_light | 2 | — | 2 | 2 / 2 | 0.931 | $18.0566 | $9.0283 |
| fable | t4_takeover | 2 | — | 2 | 0 / 2 | 0.977 | $14.1335 | $7.0667 |
| fable | all tasks | 4 | — | 4 | 2 / 4 | 0.954 | $32.1901 | $8.0475 |
| opus | t3_light | 2 | — | 2 | 1 / 2 | 0.887 | $6.9643 | $3.4821 |
| opus | t4_takeover | 2 | — | 2 | 0 / 2 | 0.967 | $4.3869 | $2.1934 |
| opus | all tasks | 4 | — | 4 | 1 / 4 | 0.927 | $11.3512 | $2.8378 |
| sol | t3_light | 3 | — | 3 | 2 / 3 | 0.921 | $1.7867 | $0.5956 |
| sol | t4_takeover | 3 | — | 3 | 2 / 3 | 0.994 | $0.9458 | $0.3153 |
| sol | all tasks | 6 | — | 6 | 4 / 6 | 0.957 | $2.7325 | $0.4554 |
| terra | t3_light | 3 | — | 3 | 2 / 3 | 0.894 | $0.8281 | $0.2760 |
| terra | t4_takeover | 3 | — | 3 | 2 / 3 | 0.997 | $0.6874 | $0.2291 |
| terra | all tasks | 6 | — | 6 | 4 / 6 | 0.946 | $1.5154 | $0.2526 |
Each trial has one end grade, computed once after the run ends. Harness failures (the run stopped on a provider or harness error) are listed but have no score: they are left out of "all pass" and the means.
| headline | argon | t3_light | two_targets_room_light | trial 0 | submitted | 0.829 | fail | $2.7663 | 55 | 122 |
| headline | argon | t3_light | two_targets_room_light | trial 1 | submitted | 0.833 | fail | $2.4947 | 47 | 107 |
| headline | argon | t3_light | two_targets_room_light | trial 2 | submitted | 0.667 | fail | $2.4138 | 41 | 56 |
| headline | astra | t3_light | two_targets_room_light | trial 0 | submitted | 0.833 | fail | $3.4831 | 50 | 81 |
| headline | astra | t3_light | two_targets_room_light | trial 1 | submitted | 0.954 | pass | $4.4147 | 52 | 92 |
| headline | astra | t3_light | two_targets_room_light | trial 2 | submitted | 0.811 | fail | $3.0374 | 51 | 77 |
| headline | fable | t3_light | two_targets_room_light | trial 0 | submitted | 0.925 | pass | $9.3349 | 48 | 118 |
| headline | fable | t3_light | two_targets_room_light | trial 1 | submitted | 0.936 | pass | $8.7217 | 56 | 116 |
| headline | opus | t3_light | two_targets_room_light | trial 0 | submitted | 0.824 | fail | $4.0500 | 59 | 133 |
| headline | opus | t3_light | two_targets_room_light | trial 1 | submitted | 0.951 | pass | $2.9143 | 57 | 98 |
| headline | sol | t3_light | two_targets_room_light | trial 0 | submitted | 0.96 | pass | $0.8340 | 65 | 120 |
| headline | sol | t3_light | two_targets_room_light | trial 1 | submitted | 0.973 | pass | $0.4585 | 52 | 84 |
| headline | sol | t3_light | two_targets_room_light | trial 2 | submitted | 0.831 | fail | $0.4942 | 52 | 89 |
| headline | terra | t3_light | two_targets_room_light | trial 0 | submitted | 0.756 | fail | $0.2301 | 43 | 72 |
| headline | terra | t3_light | two_targets_room_light | trial 1 | submitted | 0.956 | pass | $0.3710 | 57 | 89 |
| headline | terra | t3_light | two_targets_room_light | trial 2 | submitted | 0.972 | pass | $0.2270 | 38 | 69 |
| headline | argon | t4_takeover | t_present_faults | trial 0 | submitted | 0.981 | fail | $2.4025 | 20 | 68 |
| headline | argon | t4_takeover | t_present_faults | trial 1 | submitted | 0.981 | fail | $2.0029 | 15 | 45 |
| headline | argon | t4_takeover | t_present_faults | trial 2 | submitted | 0.981 | fail | $1.6851 | 17 | 46 |
| headline | astra | t4_takeover | t_present_faults | trial 0 | submitted | 0.971 | fail | $2.6635 | 30 | 48 |
| headline | astra | t4_takeover | t_present_faults | trial 1 | submitted | 1 | pass | $3.1738 | 35 | 59 |
| headline | astra | t4_takeover | t_present_faults | trial 2 | submitted | 1 | pass | $2.2606 | 30 | 46 |
| headline | fable | t4_takeover | t_present_faults | trial 0 | submitted | 0.974 | fail | $6.2998 | 15 | 66 |
| headline | fable | t4_takeover | t_present_faults | trial 1 | submitted | 0.979 | fail | $7.8337 | 19 | 67 |
| headline | opus | t4_takeover | t_present_faults | trial 0 | submitted | 0.953 | fail | $2.4213 | 20 | 64 |
| headline | opus | t4_takeover | t_present_faults | trial 1 | submitted | 0.981 | fail | $1.9656 | 19 | 57 |
| headline | sol | t4_takeover | t_present_faults | trial 0 | submitted | 1 | pass | $0.3465 | 29 | 49 |
| headline | sol | t4_takeover | t_present_faults | trial 1 | submitted | 0.981 | fail | $0.2976 | 29 | 46 |
| headline | sol | t4_takeover | t_present_faults | trial 2 | submitted | 1 | pass | $0.3017 | 29 | 48 |
| headline | terra | t4_takeover | t_present_faults | trial 0 | submitted | 1 | pass | $0.2280 | 22 | 43 |
| headline | terra | t4_takeover | t_present_faults | trial 1 | submitted | 1 | pass | $0.2257 | 23 | 47 |
| headline | terra | t4_takeover | t_present_faults | trial 2 | submitted | 0.992 | fail | $0.2337 | 24 | 52 |