GLM-5.3 · Codex · robouse-tabletop
GLM-5.3 · Codex: the model's API rejected the camera images the harness attached, so its vision trials failed as infrastructure failures and are not scored (31 of its 43; the other 12 because the episode server did not start). Its scored trials saw no images.
| Close the drawer | Solved | 134 / 400 | 1 min | |
|---|---|---|---|---|
| Gentle placement of a fragile vial | Solved | 212 / 700 | 55 s | |
| Heavy block and figureAgent gave up | Not solved | Agent gave up | 154 / 700 | 36 s |
| Heavy blocks and figureAgent gave up | Not solved | Agent gave up | 297 / 900 | 53 s |
| Insert a peg into a socketAPI rejected the image | Infra failure | API rejected the image | 63 / 600 | 20 s |
| Knife and figureAPI rejected the image | Infra failure | API rejected the image | 514 / 700 | 2 min 10 s |
| Move the blue block between the cylinder and the tray | Solved | 165 / 500 | 33 s | |
| Move the red block that is left of the bowl | Solved | 192 / 500 | 36 s | |
| Open the drawerAPI rejected the image | Infra failure | API rejected the image | 67 / 400 | 28 s |
| Open the drawer and put the block insideAPI rejected the image | Infra failure | API rejected the image | 61 / 900 | 28 s |
| Place the cylinder on the left pad | Solved | 176 / 500 | 30 s | |
| Precise landing on a small padAPI rejected the image | Infra failure | API rejected the image | 55 / 500 | 21 s |
| Push the block into the binAgent said done | Not solved | Agent said done | 697 / 700 | 3 min 8 s |
| Push the block off the tableAgent gave up | Not solved | Agent gave up | 57 / 700 | 25 s |
| Push the block onto the green pad | Solved | 181 / 500 | 39 s | |
| Put both blocks in the bowl | Solved | 307 / 800 | 1 min 16 s | |
| Put the block in the drawer and close itStep budget used up | Not solved | Step budget used up | 900 / 900 | 4 min 3 s |
| Put the green cube on the plateAPI rejected the image | Infra failure | API rejected the image | 266 / 500 | 56 s |
| Put the heavy block on the pad | Solved | 218 / 700 | 30 s | |
| Put the knife away | Solved | 218 / 700 | 51 s | |
| Put the red block in the bowl | Solved | 188 / 500 | 31 s | |
| Relocate a payload past a protected columnStep budget used up | Not solved | Step budget used up | 600 / 600 | 2 min 58 s |
| Retrieve a block from clutter | Solved | 122 / 500 | 1 min 7 s | |
| Retrieve a block from clutterAPI rejected the image | Infra failure not counted | API rejected the image | 53 / 500 | 25 s |
| Rule inference: blocks and pads | Solved | 681 / 1,200 | 33 s | |
| Rule inference: completing a patternAPI rejected the image | Infra failure | API rejected the image | 347 / 1,200 | 1 min 22 s |
| Rule inference: exchange | Solved | 693 / 1,000 | 38 s | |
| Rule inference: falling | Solved | 697 / 1,400 | 41 s | |
| Rule inference: ordering | Solved | 708 / 1,400 | 1 min 10 s | |
| Rule inference: sizesAPI rejected the image | Infra failure | API rejected the image | 334 / 800 | 48 s |
| Rule inference: sorting | Solved | 670 / 1,400 | 45 s | |
| Rule inference: the marker | Solved | 392 / 500 | 1 min 5 s | |
| Rule inference: the missing piece | Solved | 362 / 600 | 1 min 26 s | |
| Rule inference: translation | Solved | 575 / 1,400 | 36 s | |
| Rule inference: which one moves | Solved | 193 / 500 | 24 s | |
| Set up a 2x2 board position | Solved | 705 / 1,400 | 40 s | |
| Sort blocks into trays by colourAPI rejected the image | Infra failure | API rejected the image | 135 / 1,400 | 32 s |
| Stack the heavy blocksAPI rejected the image | Infra failure | API rejected the image | 247 / 900 | 38 s |
| Stack the red block on the blue block | Solved | 185 / 500 | 31 s | |
| Tower of Hanoi, three discsAgent said done | Not solved | Agent said done | 2,490 / 2,500 | 2 min 11 s |
No trials match these filters.