Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Embodied Agentic on SciWorld

88.9Accuracy

M2CL

22.65239.85157.0574.249Feb 2, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
88.9
2026.02
68.4
2026.02
65.2
2026.02
62.1
2026.02
60.2
2026.02
58
2026.02
56.1
2026.02
50.4
2026.02
46
2026.02
45.3
2026.02
41.6
2026.02
39.4
2026.02
37
2026.02
36
2026.02
35.9
2026.02
35.3
2026.02
32.2
2026.02
31.3
2026.02
30.8
2026.02
29.4
2026.02
25.2