Comparing Fable 5.1 to Fable 5 on robotic manipulation

September 3, 2026

We gave Claude Fable 5 and Claude Fable 5.1 control of a pair of YAM arms and evaluated both on two Inspect Robots manipulation tasks:

“Pick up the red block from the table and place it inside the bowl”

“Pick up the round blue puzzle piece by the knob at its center and place it into the matching circular groove in the board”

Block into bowl

Puzzle into groove

Across 20 trials per model per task, Fable 5.1 reached later stages of both tasks than Fable 5, completed each task more often, and produced fewer output tokens doing it.

One Fable 5.1 trial replayed in Rerun: the three camera views the model receives each turn, its own notes as it decides, and the end‑effector targets it commands — all on one timeline.

Fable 5.1 completes more while producing fewer tokens

Condition means. Both tasks move up and to the left from Fable 5 to Fable 5.1 — more completions for fewer output tokens. At 20 trials per condition the bowl‑task gap is clear; the puzzle task's 0 versus 2 completions is smaller, but directionally suggestive.

Scoring

Every trial was scored by a human grader on the highest stage it reached, so a run that fails still records how far it got.

Fable 5.1 gets further up the ladder on both tasks

Share of 20 trials per model reaching each stage. Fable 5 stalls at first contact in 13 of 20 bowl trials, never closing the gripper on the block. On the puzzle task both models get further, but the wall moves to the final insertion.

Results

All runs

Every counted trial, 80 in total.

Technical specifications

Limitations

• Grading was operator-judged, with the operator aware of which model was running throughout, so the scores are open to unconscious bias.
• Task objects were reset by hand after every trial, so starting configurations varied slightly.
• Both models ran only at medium thinking effort; whether the difference holds at other effort levels is untested.
• Fable 5 exhausted its 20-call budget in 18 of 20 bowl trials, where Fable 5.1 usually stopped on its own. We cannot say whether Fable 5 would close the gap given a larger call budget.