Cook Broccolini¶
Agent runs
Each mode of this task runs once per run. Est. cost is tokens at list price (data/prices.yml), never a bill; see the BEHAVIOR-1K runs for every task of a run.
Codex + GPT-6 Luna, reasoning xhigh (ChatGPT login)default run
| Trial | Started | Agent time | Model requests | Tokens in / out | Est. cost | Grade |
|---|---|---|---|---|---|---|
| U✗ 4h 00m 36% | 09-27 07:40 | 4h 00m | 551 | 77.1M / 227k | $0.978 | deterministic 1, grader_error 0, instance_exact 0, instance_ok 1, n_actions 5657, video_rendered 1 |
| L✗ 4h 00m 0% | 09-27 10:07 | 4h 00m | 864 | 121.2M / 482k | $1.63 | deterministic 1, grader_error 0, live_success 0, n_actions 11767, replay_success 0, video_rendered 1 |
No instruction published upstream
The 2025 carryover tasks ship without instruction text. Watch the demo and write the goal in your own words below, clearly marked as a reconstruction rather than the official goal.
Measured on the stack we run — BEHAVIOR-1K v3.9.2 / OmniGibson 3.9.2 / Isaac Sim 5.1
| Goal predicates | cooked ontop |
| Quantifiers | forall |
| Goal clauses | 3 |
| Objects named in the goal | broccolini.n.01 clove.n.03 frying_pan.n.01 stove.n.01 |
| Objects in the problem | 11 in 8 categories |
| Rooms loaded | kitchen_0 |
| Demonstrations | 200 teleoperated episodes |
| Mean episode | 2m 34s (4,618 control steps at 30 Hz) |
| Base travel, mean | 10.87 m |
| Gripper travel, mean | left 5.91 m · right 6.31 m |
| Evaluation instances | 20 public test instances (ids 301–320) |
Read from BDDL activity definitions + 2026-challenge-task-instances (licensed download), 2026-09-21.
Tags¶
Task DomainMobile / Whole-body Manipulation
Why this task is interesting¶
Not yet written.
Capability notes¶
Not yet written.
Oracle demo review¶
Not yet reviewed.