# Apple paper finds a general AI model controlling a robot arm beats specialist policies

> Original title: New Apple paper shows that a general-purpose AI model like Claude Opus 5.5, given camera views and direct control of a robot arm, handled...

- Source: Rohan Paul
- Published: 2026-11-09T23:43:42.000Z
- Bad Influence: https://badinfluence.news/items/l3p9p3acfw6xed75812eu2ia7
- Original: https://x.com/rohanpaul_ai/status/2119939013935255693

## Summary

Apple researchers report that AMP1, a harness letting a general AI model like Claude Opus 5.5 control a robot arm directly, succeeded in 85.6% of RoboLab-120 episodes zero-shot. That compares with 49.2% for the best vision-language-action model and 42.9% for the best world-action model. On a real Franka arm, the system succeeded in 104 of 108 trials, though the robot waits much of each episode for the model.

## Why it matters

The paper reports zero-shot robot results against specialist policies, with a clear harness design and a real-arm trial count that show the approach's limits.

## Post

New Apple paper shows that a general-purpose AI model like Claude Opus 5.5, given camera views and direct control of a robot arm, handled objects more reliably than robot models trained on recorded demonstrations.

New Apple paper shows that a frontier AI model directly controlling a robot arm beats specialist robot policies, with no demonstrations or task training.

Most robot manipulation work trains specialist policies, like vision-language-action models, on demonstrations, and they break easily when instructions change.

Their harness, AMP1, gives a model like Claude Opus 5.5 or Gemini 4 Argon camera views, code execution, and direct motion commands in a loop. It has no built-in grasping or placing skills, so the model invents its own strategies.

On RoboLab-120, AMP1 succeeded in 85.6% of episodes zero-shot, against 49.2% for the best VLA and 42.9% for the best world-action model. On a real Franka arm it succeeded in 104 of 108 trials, though the robot waits much of each episode for the model.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWuJR3yb0AAsPkW.png&mode=thumb&exp=1794528000&sig=3ee0c3430e88ec0c)