Embodied Question Answering
Benchmarks
DatasetSOTA methodMetricTrendResultsLast updated
85.1Overall (LLM-Match)
33
Jun 24, 2026
67.7LLM-Match
32
Feb 26, 2026
69.3Accuracy
26
Jun 11, 2026
86.8Accuracy
24
Jul 3, 2026
86.8LLM Match
19
Apr 7, 2026
64.8Answer Correctness
15
Apr 21, 2026
88.9Image Citation Relevance
15
Apr 21, 2026
3.91QAS (Score)
14
Jun 16, 2026
0.15Steps
14
Jun 17, 2026
77.1BLEU-1
13
Feb 26, 2026
93.02Existence Accuracy
12
Jul 3, 2026
58.3LLM-Match
10
Apr 3, 2026
86.8LLM-Match Score
10
Mar 25, 2026
0.19Steps
10
Jun 17, 2026
86.8GPT-4 Score
10
Mar 31, 2026
80Exact Match
10
Feb 26, 2026
86.8LLM-Match
8
Feb 26, 2026
75.41Total Accuracy
7
Jun 30, 2026
52.6LLM-Match
7
Feb 26, 2026
9.97Response Latency (s)
6
Apr 21, 2026
95.35EQA Accuracy
6
Apr 21, 2026
55.9LLM-Match
6
Apr 7, 2026
79.27LLM Score
6
Jun 17, 2026
56.34Accuracy
6
Feb 26, 2026
58.2Accuracy
5
Mar 11, 2026