Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Document Question Answering on CharXiv RQ
Loading...
68.6
Accuracy
GPT-5 mini
48.112
53.431
58.75
64.069
May 28, 2026
Accuracy
Updated 5d ago
Evaluation Results
Method
Method
Links
Accuracy
GPT-5 mini
Version=high
2026.05
68.6
RLR³
Training Source=OpenMMR
2026.05
62.7
RLVR
Training Source=ViRL
2026.05
62.4
RLR³
Training Source=DeepVi...
2026.05
61.8
RLR³
Training Source=ViRL
2026.05
61.4
RLVR
Training Source=DeepVi...
2026.05
61.2
RLVR
Training Source=OpenMMR
2026.05
58.4
Official thinking
Source=Qwen3-VL
2026.05
56.6
Base instruct
2026.05
53.4
Official instruct
Source=Qwen3-VL
2026.05
48.9
GPT-5 mini
Version=minimal
2026.05
48.9
Feedback
Search any
task
Search any
task