Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Active Reasoning on ARBench-GN
Loading...
45.3
Accuracy
TeamTR (Homogeneous)
38.644
40.372
42.1
43.828
May 1, 2026
Accuracy
Updated 2mo ago
Evaluation Results
Method
Method
Links
Accuracy
TeamTR (Homogeneous)
Size=3×8B
2026.05
45.3
o4-mini
Size=–
2026.05
45.1
TeamTR (Heterogeneous)
Size=1.7B+8B+14B
2026.05
44.5
GPT-4o-mini
Size=–
2026.05
43.6
Debate + Judge
Size=3×8B
2026.05
43.5
Role-play
Size=3×8B
2026.05
43.1
Qwen3-A3B
Size=30B
2026.05
42.3
Debate
Size=3×8B
2026.05
42.3
Qwen3 (thinking)
Size=32B
2026.05
41.1
DeepSeek-R1 Distill
Size=70B
2026.05
40.7
QwQ
Size=32B
2026.05
39.5
DAPO
Size=8B
2026.05
39.3
GRPO
Size=8B
2026.05
38.9
Feedback
Search any
task
Search any
task