EmbodiedLab Fine Reading (T0 success)
Agents · 2026-10-26
The seven EmbodiedLab experiments whose primary capability dimension is fine reading of scales, displays and positions (six basic, one advanced), such as Hooke's law, a lens focal length and single-slit diffraction. Same T0 interface, scoring gates and averaging as the full benchmark; the score is the success rate on these experiments, in percent.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 42.9 |
| GPT-6 Astra | 42.9 |
| GPT-6.1 Sol | 35.7 |
| DeepSeek V4.1 Flash 1017 | 14.3 |
| Kimi K3 | 14.3 |
| GLM-5.3 Flash | 0.0 |