Red brush · sweep forward¶
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 DexToolBench 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✗ 37m 42% | 10-01 06:45 | 37m | — | 12.8M / 178k | $0.252 | deterministic 1, grader_error 0, n_actions 5092, video_rendered 1 |
| L✗ 59m 0% | 10-01 07:17 | 59m | — | 22.2M / 203k | $0.353 | deterministic 1, grader_error 0, video_rendered 1 |
1 other run(s) of DexToolBench
Codex + GPT-6 Luna, reasoning xhigh (Azure OpenAI API, 2026-09-30 stress test)
| Trial | Started | Agent time | Model requests | Tokens in / out | Est. cost | Grade |
|---|---|---|---|---|---|---|
| U✗ 1h 00m 0% | 09-30 08:04 | 1h 00m | — | 11.3M / 170k | $0.229 | grader_error 0, missing_trajectory 1 |
| L✗ 58m 0% | 09-30 09:53 | 58m | — | 13.7M / 88k | $0.194 | deterministic 1, grader_error 0, video_rendered 1 |
Task instruction (upstream)
Pick up the red brush and move it through the demonstrated motion (a forward sweep): bring it to each of the 38 goal poses in order.
Recorded by us: SimToolReal's pretrained RL policy replayed in our MuJoCo port of the task, four views (front, side / top, oblique); it reaches 20 of 38 goals here. The green ghost tool is the current goal.
What DexToolBench states about this task
| Defined in | dextoolbench/trajectories/brush/red_brush/sweep_forward.json |
| Category | brush |
| Object | red_brush |
| Task | sweep_forward |
| N Goals | 38 |
| Tool Model | assets/urdf/dextoolbench/brush/red_brush/red_brush.urdf |
| Table | table_narrow.urdf (plain) |
| Metric | task progress: goals reached / goals (8 grasp-box keypoints within 1.5 cm; 10 s per goal) |
From https://github.com/tylerlum/simtoolreal @ 313d5ae.
Measured on the stack we run — MuJoCo 3.3.7 (CPU, bit-exact replay), KUKA iiwa 14 (MuJoCo Menagerie 1b86ece) + Sharpa HA4, 600 Hz physics, 60 Hz control
| Harness Task | dextoolbench_brush_red_brush_sweep_forward |
| Oracle | SimToolReal's pretrained policy, recorded in this scene and replayed: 20/38 goals |
Read from run in our MuJoCo port (robot_coding_bench tasks/dextoolbench_brush_red_brush_sweep_forward, image rcb-mujoco 0.1.1), 2026-09-28.
Tags¶
Why this task is interesting¶
Not yet written.
Capability notes¶
Not yet written.
Oracle demo review¶
Not yet reviewed.